Fine-Grained Sports, Yoga, and Dance Postures Recognition: A Benchmark Analysis (2308.00323v1)

Abstract: Human body-pose estimation is a complex problem in computer vision. Recent research interests have been widened specifically on the Sports, Yoga, and Dance (SYD) postures for maintaining health conditions. The SYD pose categories are regarded as a fine-grained image classification task due to the complex movement of body parts. Deep Convolutional Neural Networks (CNNs) have attained significantly improved performance in solving various human body-pose estimation problems. Though decent progress has been achieved in yoga postures recognition using deep learning techniques, fine-grained sports, and dance recognition necessitates ample research attention. However, no benchmark public image dataset with sufficient inter-class and intra-class variations is available yet to address sports and dance postures classification. To solve this limitation, we have proposed two image datasets, one for 102 sport categories and another for 12 dance styles. Two public datasets, Yoga-82 which contains 82 classes and Yoga-107 represents 107 classes are collected for yoga postures. These four SYD datasets are experimented with the proposed deep model, SYD-Net, which integrates a patch-based attention (PbA) mechanism on top of standard backbone CNNs. The PbA module leverages the self-attention mechanism that learns contextual information from a set of uniform and multi-scale patches and emphasizes discriminative features to understand the semantic correlation among patches. Moreover, random erasing data augmentation is applied to improve performance. The proposed SYD-Net has achieved state-of-the-art accuracy on Yoga-82 using five base CNNs. SYD-Net's accuracy on other datasets is remarkable, implying its efficiency. Our Sports-102 and Dance-12 datasets are publicly available at https://sites.google.com/view/syd-net/home.

• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state-of-the-art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.
• The paper introduces SYD-Net, a deep learning model incorporating a patch-based attention mechanism to improve fine-grained recognition of sports, yoga, and dance postures.
• The paper contributes two new benchmark datasets, Sports-102 and Dance-12, addressing the need for more diverse image data for fine-grained posture recognition.
• The proposed SYD-Net achieves state_of_the_art accuracy on existing Yoga-82 and Yoga-107 datasets and demonstrates improved performance on the new datasets.

The paper addresses the problem of fine-grained image classification (FGIC) for Sports, Yoga, and Dance (SYD) postures, which is challenging due to complex body movements and large intra-class variations. To tackle the limitations of existing datasets, the authors introduce two new image datasets: Sports-102, comprising 102 sport categories, and Dance-12, consisting of 12 dance styles. They also experiment with two public yoga datasets, Yoga-82 and Yoga-107.

The authors propose a deep learning model, SYD-Net, which incorporates a patch-based attention (PbA) mechanism on top of standard backbone convolutional neural networks (CNNs). The PbA module utilizes a self-attention mechanism to learn contextual information from uniform and multi-scale patches, emphasizing discriminative features to understand semantic correlations. The architecture of SYD-Net is shown in Figure \ref{fig:Model2a}. Random erasing data augmentation is also employed to enhance performance.

The key contributions include:

• A patch-based attention mechanism for fine-grained SYD posture recognition.
• Two new image datasets for sports and dance action classification without part-based/skeletal-joint/bounding-box information.
• Extensive experiments using five backbone CNN architectures on the four SYD datasets.
• State-of-the-art accuracy on the Yoga-82 and Yoga-107 datasets.

Datasets

The paper describes the creation and utilization of four image datasets for SYD posture recognition. A summary of dataset details is shown in Table \ref{DB}.

• Sports-102: This dataset contains images of 102 sport action categories, addressing the lack of image-based sports datasets. Examples from the dataset can be seen in Figure \ref{fig:s102}, and class distributions are visualized in Figure \ref{fig:data_Sports}.
• Dance-12: The Dance-12 dataset is a new image dataset consisting of 12 dance styles with examples in Figure \ref{fig:DanceStyles}. The distribution of training and testing splits is shown in Figure \ref{fig:data_Dance}.
• Yoga-82: This dataset, comprised of 82 fine-grained yoga poses, was introduced in a previous work. The version used in this paper consists of 19.9k training and 7.2k testing images after standardization. Samples are shown in Figure \ref{fig:yoga82_samples}.
• Yoga-107: Collected from Kaggle, the Yoga-107 dataset includes 107 fine-grained yoga pose classes, totaling 5.9k images, as shown in Figure \ref{fig:yoga82_samples}.

Method

The SYD-Net model integrates spatial and channel attention mechanisms on top of standard backbone CNNs and consists of three main modules: uniform and multi-scale patch proposals, a patch-based attention module, and classification.

1. Uniform and Multi-Scale Patch Proposals:
   • The input image $I_l \in \mathbb{R}^{h\times w\times 3}$ is fed into a backbone CNN $\mathcal{N}$ to extract high-level feature maps $F \in \mathbb{R}^{h\times w\times c}$.
   • The image is divided into a set $D$ of $e$ non-overlapping uniform patch proposals $d_i$ with spatial size $a \times a$, where $e = \frac{h \times w}{a^2}$.
   • Multi-scale patches are defined to capture hierarchical information with varying patch sizes $p_i = [x_i, y_i, \Delta w_i, \Delta h_i]$ and $p_j = [x_j, y_j, \Delta w_j, \Delta h_j]$ such that $\Delta w_i > \Delta w_j$ and $\Delta h_i > \Delta h_j$.
   • The collection of all $n$ patches (uniform and multi-scale) is denoted as $P = \{p_i\}_{i=1}^{i=n}$, and the corresponding feature map is $$F = \{F_i\}_{i=1}^{i=n} \in \mathbb{R}^{n \times (h \times w \times c)}$$.
   • $F$ is upsampled to a higher resolution $k(h \times w)$, and bilinear interpolation is used for pooling features from every patch, maintaining the feature vector sizes as $F \in \mathbb{R}^{(h \times w \times c)}$.
2. Patch-based Attention (PbA) Mechanism:

   • Channel Attention (CA):

     • Self-attention is applied to capture channel-wise relationships, using query Q, key K, and value V learned from the input feature vector F.

     • Attentional feature map:

       $\psi_{i,j} = tanh({\mathbf{W}_\psi F_i} + {\mathbf{W}_{\psi'}F_{j}+\mathbf{b}_{\psi})$

       $$\vartheta_{i,j} = \sigma\left(\mathbf{W}_\vartheta \psi_{i,j} + \mathbf{b}_{\vartheta} \right)$$

       where:

       • $F_i$ and $F_j$ are high-level feature vectors from patches $p_i$ and $p_j$.
       • $W_\psi$ and $W_{\psi'}$ are weight matrices for computing attention vectors.
       • $W_\vartheta$ is a weight matrix for nonlinear combination.
       • $b_\psi$ and $b_\vartheta$ are bias vectors.
       • $\sigma$ is a nonlinear activation function.
     • The importance of each patch $p_i$ is computed as a weighted sum of attention scores:

       $$\delta_{i,j} = softmax(\mathbf{W}_{\delta}\vartheta_{i,j}+\mathbf{b}_\delta)$$

       $$\hat{\mathbf{F}_i} = \sum_{j=1}^{n}\delta_{i,j}F_{j}$$

       where:

       • $W_\delta$ is a weight matrix.
       • $b_\delta$ is a bias vector.
     • The aggregated feature space $\hat{F}_i$ is summarized using global average pooling (GAP) to generate $\tilde{\mathbf{F}_i}$, which is passed through a softmax layer to produce a weighted attention matrix $\phi_{i}$.
     • The output of the cross-channel attention mechanism is:

       $${\tilde{\mathbf{F}_i}} = \mathcal{GAP}\left({\hat{\mathbf{F}_i}}\right)$$

       $${\mathbf{F}_{CA}} = \sum_{i=1}^{n}\phi_{i}{\tilde{\mathbf{F}_i}}$$

       $$\phi_{i} = softmax(\mathbf{W}_{\phi}{\tilde{\mathbf{F}_i}+\mathbf{b}_\phi})$$

* Spatial Attention (SA): * Spatial attention captures neighborhood information to refine global structural information by generating an attentional mask. * Global average pooling (GAP) and global max pooling (GMP) are applied to refine spatial features $F$ from all patches $P$, generating $F_{gap}$ and $F_{gmp}$. * The fused feature map $H \in \mathbb{R}^{n(h \times w \times 2)}$ is defined as:

$$H = concat\Bigl(\mathcal{GAP}\left( F \right); \mathcal{GMP}\left( F \right) \Bigl)$$

* A multi-layer perceptron (MLP) is applied to generate a spatial attention mask $F_{SA}$:

$F_{SA} = \mathcal{MLP} \big(H \big)$

* The spatial attention mask $F_{SA} \in \mathbb{R}^{(1 \times c)}$ is multiplied element-wise with the weighted attention vector $F_{CA}$ from the channel attention method. * A residual path is connected with $F_{CA}$ for smoother gradient flow. * The patch-based attention (PbA) feature vector $F_{PbA} \in \mathbb{R}^{(1 \times c)}$ is obtained as:

$$F_{PbA} = \Bigl(F_{CA} \otimes F_{SA} + F_{CA}\Bigl)$$

1. Classification:
   • The upsampled feature $F$ is squeezed by a GAP layer to produce a vector of $(1 \times c)$ channels, which is added to the attentional feature map $F_{PbA}$.

   • The final feature vector $F_{f}$ is regularized and passed to a softmax layer to compute class probabilities:

     $F_{f} = F_{PbA}+ \mathcal{GAP}\left(F \right)$

     $Y_{pred}= softmax(\lambda(F_{f}) )$

     where $\lambda$ denotes a regularization process with Gaussian dropout (GD) and batch normalization (BN).

Experiments

• Implementation Details:
  • The model is implemented using ResNet-50, DenseNet-201, NASNetMobile, MobileNet-v2, and Xception backbone CNNs in TensorFlow-2.x.
  • The input image resolution is 224$\times$224.
  • Three sets of uniform patches (3$\times$3, 4$\times$4, and 5$\times$5) and corresponding hierarchical regions are generated.
  • Standard data augmentations of random rotation ($\pm$25 degrees) and random scaling (1$\pm$0.25) are applied. Two random regions are erased with either a fixed RGB=127 or random RGB pixel values.
  • SYD-Net is trained from scratch and with ImageNet weights.
  • Stochastic Gradient Descent (SGD) is used with an initial learning rate of $0.007$, multiplied by 0.1 after every 50 epochs.
  • The model is trained for 200 epochs with a mini-batch size of 8 using a Tesla M10 GPU of 8 GB.
  • Gaussian dropout rate 0.2 and batch normalization are applied.
• Baseline Results:
  • Baseline performances are computed using four backbone CNNs (trained from scratch) on the four datasets, as shown in Table \ref{baseline}.
  • Three sets of experiments are conducted: conventional data augmentation, conventional augmentation with random region erasing, and attention with random erasing.
• Ablation Studies:
  • The performances of a different number of patches using four base CNNs are evaluated on all four datasets, as shown in Table \ref{table:RoI_Acc}.
  • The impact of incorporating uniform and hierarchical patches is analyzed.
  • The top-5 accuracy (\%) of SYD-Net using $P_{30}$ is measured, as shown in Table \ref{top-5}.
  • The effect of training with pre-trained ImageNet weights is evaluated.
• Performance Comparison:
  • The model's complexity is evaluated with the number of parameters, and the performance is compared with state-of-the-art methods.
  • The top-1 accuracies using Xception, trained from scratch and with ImageNet weights, are reported, as shown in Table \ref{DB}.
  • A comparative paper on Yoga-107 using five base CNNs is conducted, as illustrated in Figure \ref{fig:CompY107}.
  • The performances on Yoga-82 using various CNNs are compared to existing methods, as detailed in Table \ref{Comparson_Y82}.

Results and Discussion

• The patch-based attention mechanism improves accuracy by integrating spatial and channel attention.
• Hybrid patches (uniform and hierarchical) enhance performance by capturing contextual information at multiple granularities.
• Random region erasing data augmentation contributes to better generalization and performance.
• MobileNet-v2 and NASNetMobile achieve competitive results with lower model complexity.
• SYD-Net outperforms existing methods on the Yoga-82 and Yoga-107 datasets, achieving state-of-the-art results.
• Visualizations of feature maps using t-SNE plots demonstrate improved class separability with SYD-Net. Confusion matrices of Dance-12 are shown in Figure \ref{fig:ConMat}, and t-SNE plots using MobileNet-v2 are in Figure \ref{fig:tSNE}.

The paper concludes by highlighting the effectiveness of the proposed SYD-Net for fine-grained human posture recognition and introducing the new Sports-102 and Dance-12 datasets for future research.