Papers
Topics
Authors
Recent
Gemini 2.5 Flash
Gemini 2.5 Flash
110 tokens/sec
GPT-4o
56 tokens/sec
Gemini 2.5 Pro Pro
44 tokens/sec
o3 Pro
6 tokens/sec
GPT-4.1 Pro
47 tokens/sec
DeepSeek R1 via Azure Pro
28 tokens/sec
2000 character limit reached

Deep Learning-Based Human Pose Estimation: A Survey (2012.13392v5)

Published 24 Dec 2020 in cs.CV, cs.GR, and cs.MM

Abstract: Human pose estimation aims to locate the human body parts and build human body representation (e.g., body skeleton) from input data such as images and videos. It has drawn increasing attention during the past decade and has been utilized in a wide range of applications including human-computer interaction, motion analysis, augmented reality, and virtual reality. Although the recently developed deep learning-based solutions have achieved high performance in human pose estimation, there still remain challenges due to insufficient training data, depth ambiguities, and occlusion. The goal of this survey paper is to provide a comprehensive review of recent deep learning-based solutions for both 2D and 3D pose estimation via a systematic analysis and comparison of these solutions based on their input data and inference procedures. More than 250 research papers since 2014 are covered in this survey. Furthermore, 2D and 3D human pose estimation datasets and evaluation metrics are included. Quantitative performance comparisons of the reviewed methods on popular datasets are summarized and discussed. Finally, the challenges involved, applications, and future research directions are concluded. A regularly updated project page is provided: \url{https://github.com/zczcwh/DL-HPE}

Definition Search Book Streamline Icon: https://streamlinehq.com
References (316)
  1. PoseTrack: A Benchmark for Human Pose Estimation and Tracking. In CVPR.
  2. Posetrack: A benchmark for human pose estimation and tracking. In CVPR.
  3. 2d human pose estimation: New benchmark and state of the art analysis. In CVPR.
  4. Actionxpose: A novel 2d multi-view pose-based algorithm for real-time human action recognition. In arXiv preprint arXiv:1810.12126.
  5. Exploiting Temporal Context for 3D Human Pose Estimation in the Wild. In CVPR.
  6. Bruno Artacho and Andreas Savakis. 2020. UniPose: Unified Human Pose Estimation in Single Images and Videos. In CVPR.
  7. Vasileios Belagiannis and Andrew Zisserman. 2017. Recurrent human pose estimation. In FG.
  8. PandaNet: Anchor-Based Single-Shot Multi-Person 3D Pose Estimation. In CVPR.
  9. Learning temporal pose estimation from sparsely-labeled videos. In NeurIPS.
  10. Keep it SMPL: Automatic Estimation of 3D Human Pose and Shape from a Single Image. In ECCV.
  11. Adrian Bulat and Georgios Tzimiropoulos. 2016. Human pose estimation via convolutional part heatmap regression. In ECCV.
  12. 3D Pictorial Structures for Multiple View Articulated Pose Estimation. In CVPR.
  13. Exploiting Spatial-Temporal Relationships for 3D Pose Estimation via Graph Convolutional Networks. In ICCV.
  14. Learning Delicate Local Representations for Multi-Person Pose Estimation. In arXiv preprint arXiv:2003.04030.
  15. Long-term human motion prediction with scene context. In ECCV.
  16. Realtime multi-person 2d pose estimation using part affinity fields. In CVPR.
  17. Human pose estimation with iterative error feedback. In CVPR.
  18. Ching-Hang Chen and Deva Ramanan. 2017. 3D Human Pose Estimation = 2D Pose Estimation + Matching. In CVPR.
  19. Unsupervised 3D Pose Estimation With Geometric Self-Supervision. In CVPR.
  20. Multi-person 3D Pose Estimation in Crowded Scenes Based on Multi-View Geometry. In ECCV.
  21. Patient-specific pose estimation in clinical environments. In JTEHM, Vol. 6. 1–11.
  22. Cross-View Tracking for Multi-Human 3D Pose Estimation at Over 100 FPS. In CVPR.
  23. Anatomy-aware 3D Human Pose Estimation with Bone-based Pose Decomposition. In TCSVT, Vol. 32. 198–209.
  24. A simple framework for contrastive learning of visual representations. In arXiv preprint arXiv:2002.05709.
  25. Fall Detection Based on Key Points of Human-Skeleton Using OpenPose. In Symmetry, Vol. 12. 744.
  26. Xianjie Chen and Alan L Yuille. 2014. Articulated pose estimation by a graphical model with image dependent pairwise relations. In NeurIPS.
  27. Adversarial posenet: A structure-aware convolutional network for human pose estimation. In ICCV.
  28. Monocular human pose estimation: A survey of deep learning-based methods. In CVIU, Vol. 192. 102897.
  29. Cascaded pyramid network for multi-person pose estimation. In CVPR.
  30. Towards Part-aware Monocular 3D Human Pose Estimation: An Architecture Search Approach. In ECCV.
  31. HigherHRNet: Scale-Aware Representation Learning for Bottom-Up Human Pose Estimation. In CVPR.
  32. Graph and Temporal Convolutional Networks for 3D Multi-person Pose Estimation in Monocular Videos. In AAAI.
  33. Monocular 3D Multi-Person Pose Estimation by Integrating Top-Down and Bottom-Up Networks. In CVPR.
  34. Occlusion-Aware Networks for 3D Human Pose Estimation in Video. In ICCV.
  35. Cross-Attention of Disentangled Modalities for 3D Human Mesh Recovery with Transformers. In ECCV.
  36. Beyond Static Features for Temporally Consistent 3D Human Pose and Shape From a Video. In CVPR.
  37. Pose2Mesh: Graph Convolutional Network for 3D Human Pose and Mesh Recovery from a 2D Human Pose. In ECCV.
  38. Self adversarial training for human pose estimation. In APSIPA ASC.
  39. Structured feature learning for pose estimation. In CVPR.
  40. Multi-context attention for human pose estimation. In CVPR.
  41. Optimizing Network Structure for 3D Human Pose Estimation. In ICCV.
  42. Bodies at Rest: 3D Human Pose and Shape Estimation From a Pressure Image Using Synthetic Data. In CVPR.
  43. Learning 3D Human Pose from Structure and Motion. In ECCV.
  44. VPN: Learning Video-Pose Embedding for Activities of Daily Living. In ECCV.
  45. Adapting MobileNets for mobile based upper body pose estimation. In AVSS.
  46. Joint flow: Temporal flow fields for multi person tracking. In arXiv preprint arXiv:1805.04596.
  47. Fast and Robust Multi-Person 3D Pose Estimation From Multiple Views. In CVPR.
  48. Shape-aware Multi-Person Pose Estimation from Multi-view Images. In ICCV.
  49. Can 3d pose be learned from 2d projections alone?. In ECCV.
  50. 2d articulated human pose estimation and retrieval in (almost) unconstrained still images. In IJCV, Vol. 99. 190–214.
  51. Neural Architecture Search: A Survey. In JMLR, Vol. 20. 1997–2017.
  52. Compressed Volumetric Heatmaps for Multi-Person 3D Pose Estimation. In CVPR.
  53. Combining local appearance and holistic view: Dual-source deep neural networks for human pose estimation. In CVPR.
  54. Rmpe: Regional multi-person pose estimation. In ICCV.
  55. Learning to refine human pose estimation. In CVPR Workshops.
  56. Martin Fisch and Ronald Clark. 2020. Orientation Keypoints for 6D Human Pose Estimation. In arXiv preprint arXiv:2009.04930.
  57. Hierarchical Kinematic Human Mesh Recovery. In ECCV.
  58. Chained predictions using convolutional neural networks. In ECCV.
  59. Human pose estimation from monocular images: A comprehensive survey. In Sensors, Vol. 16. 1966.
  60. Generative adversarial nets. In NeurIPS.
  61. Home-based physical therapy with an interactive computer vision system. In ICCV Workshops.
  62. Multi-domain pose network for multi-person pose estimation and tracking. In ECCV Workshops.
  63. In the Wild Human Pose Estimation Using Explicit 2D Features and Intermediate 3D Representations. In CVPR.
  64. Resolving 3D Human Pose Ambiguities with 3D Scene Constraints. In ICCV.
  65. Deep residual learning for image recognition. In CVPR.
  66. Human pose estimation and activity recognition from multi-view videos: Comparative explorations of recent developments. In IEEE J Sel Top Signal Process, Vol. 6. 538–552.
  67. Part Aware Contrastive Learning for Self-Supervised Action Recognition. In IJCAI.
  68. End-to-end Dynamic Matching Network for Multi-view Multi-person 3d Pose Estimation. In ECCV.
  69. DeepFuse: An IMU-Aware Network for Real-Time 3D Human Pose Estimation from Multi-View Image. In WACV.
  70. The Devil is in the Details: Delving into Unbiased Data Processing for Human Pose Estimation. In CVPR.
  71. A coarse-fine network for keypoint localization. In ICCV.
  72. Invariant Representation Learning for Infant Pose Estimation with Small Data. In arXiv preprint arXiv:2010.06100.
  73. Deep Inertial Poser: Learning to Reconstruct Human Pose from Sparse Inertial Measurements in Real Time. In ACM TOG.
  74. Arttrack: Articulated multi-person tracking in the wild. In CVPR.
  75. Deepercut: A deeper, stronger, and faster multi-person pose estimation model. In ECCV.
  76. Human3.6M: Large Scale Datasets and Predictive Methods for 3D Human Sensing in Natural Environments. In IEEE TPAMI, Vol. 36. 1325–1339.
  77. Umar Iqbal and Juergen Gall. 2016. Multi-person pose estimation with local joint-to-person associations. In ECCV.
  78. Optical Non-Line-of-Sight Physics-Based 3D Human Pose Estimation. In CVPR.
  79. Ehsan Jahangiri and Alan L Yuille. 2017. Generating multiple diverse hypotheses for human 3d pose consistent with 2d joint detections. In ICCV Workshops.
  80. Modeep: A deep learning framework using motion features for human pose estimation. In ACCV.
  81. On the Robustness of Human Pose Estimation. In CVPR Workshops.
  82. Towards understanding action recognition. In ICCV.
  83. A survey on monocular 3D human pose estimation. In Virtual Reality &\&& Intelligent Hardware, Vol. 2. 471–500.
  84. Xiaofei Ji and Honghai Liu. 2009. Advances in view-invariant human motion analysis: a review. In IEEE TSMC, Vol. 40. 13–24.
  85. Skeleton-Aware 3D Human Shape Reconstruction From Point Clouds. In ICCV.
  86. Coherent Reconstruction of Multiple Humans From a Single Image. In CVPR.
  87. Differentiable Hierarchical Graph Grouping for Multi-Person Pose Estimation. In arXiv preprint arXiv:2007.11864.
  88. Whole-Body Human Pose Estimation in the Wild. In arXiv preprint arXiv:2007.11858.
  89. Sam Johnson and Mark Everingham. 2010. Clustered Pose and Nonlinear Appearance Models for Human Pose Estimation. In BMVC.
  90. Sam Johnson and Mark Everingham. 2011. Learning effective human pose estimation from inaccurate annotation. In CVPR.
  91. Panoptic Studio: A Massively Multiview System for Social Interaction Capture. In IEEE TPAMI. 3334–3342.
  92. Total Capture: A 3D Deformation Model for Tracking Faces, Hands, and Bodies. In CVPR.
  93. A Multi-view RGB-D Approach for Human Pose Estimation in Operating Rooms. In WACV.
  94. Ladislav Kavan. 2014. Part I: direct skinning methods and deformation primitives. In ACM SIGGRAPH.
  95. Multi-scale structure-aware network for human pose estimation. In ECCV.
  96. Transformers in vision: A survey. In arXiv preprint arXiv:2101.01169.
  97. VIBE: Video inference for human body pose and shape estimation. In CVPR.
  98. Multiposenet: Fast multi-person pose estimation using pose residual network. In ECCV.
  99. Self-Supervised Learning of 3D Human Pose Using Multi-View Geometry. In CVPR.
  100. Learning to Reconstruct 3D Human Pose and Shape via Model-Fitting in the Loop. In ICCV.
  101. Convolutional Mesh Regression for Single-Image Human Shape Reconstruction. In CVPR.
  102. Probabilistic Modeling for Human Mesh Recovery. In ICCV.
  103. Pifpaf: Composite fields for human pose estimation. In CVPR.
  104. Imagenet classification with deep convolutional neural networks. In NeurIPS.
  105. Unsupervised Cross-Modal Alignment for Multi-Person 3D Pose Estimation. In ECCV.
  106. Self-Supervised 3D Human Pose Estimation via Part Guided Novel Image Synthesis. In CVPR.
  107. Kinematic-Structure-Preserved Representation for Unsupervised 3D Human Pose Estimation. In AAAI.
  108. Unite the People: Closing the Loop Between 3D and 2D Human Representations. In CVPR.
  109. Propagating LSTM: 3D Pose Estimation based on Joint Interdependency. In ECCV.
  110. Chen Li and Gim Hee Lee. 2019. Generating Multiple Hypotheses for 3D Human Pose Estimation With Mixture Density Network. In CVPR.
  111. Human pose regression with residual log-likelihood estimation. In ICCV.
  112. HMOR: Hierarchical Multi-Person Ordinal Relations for Monocular Multi-Person 3D Pose Estimation. In ECCV.
  113. Crowdpose: Efficient crowded scenes pose estimation and a new benchmark. In ICCV.
  114. HybrIK: A Hybrid Analytical-Neural Inverse Kinematics Solution for 3D Human Pose and Shape Estimation. In CVPR.
  115. Pose Recognition with Cascade Transformers. In CVPR.
  116. Sijin Li and Antoni B. Chan. 2014. 3D Human Pose Estimation from Monocular Images with Deep Convolutional Neural Network. In ACCV.
  117. Heterogeneous multi-task learning for human pose estimation with deep convolutional neural network. In CVPR Workshops.
  118. Maximum-Margin Structured Learning With Deep Networks for 3D Human Pose Estimation. In ICCV.
  119. MHFormer: Multi-Hypothesis Transformer for 3D Human Pose Estimation. In CVPR.
  120. Rethinking on multi-stage networks for human pose estimation. In arXiv preprint arXiv:1901.00148.
  121. Test-time personalization with a transformer for human pose estimation. In NeurIPS.
  122. Dense Intrinsic Appearance Flow for Human Pose Transfer. In CVPR.
  123. SimCC: A Simple Coordinate Classification Perspective for Human Pose Estimation. In ECCV.
  124. TokenPose: Learning Keypoint Tokens for Human Pose Estimation. In ICCV.
  125. A novel joint points and silhouette-based method to estimate 3D human pose and shape. In arXiv preprint arXiv:2012.06109.
  126. On Boosting Single-Frame 3D Human Pose Estimation via Monocular Videos. In ICCV.
  127. Junbang Liang and Ming C. Lin. 2019. Shape-Aware Human Pose and Shape Reconstruction Using Multi-View Images. In ICCV.
  128. Human pose estimation using deep consensus voting. In ECCV.
  129. End-to-end human pose and mesh reconstruction with transformers. In CVPR.
  130. Mesh Graphormer. In ICCV.
  131. Microsoft coco: Common objects in context. In ECCV.
  132. Human in events: A large-scale benchmark for human-centric video analysis in complex events. In arXiv preprint arXiv:2005.04490.
  133. Polarized self-attention: towards high-quality pixel-wise regression. In arXiv preprint arXiv:2107.00782.
  134. Adversarial Attack on Skeleton-based Human Action Recognition. In arXiv preprint arXiv:1909.06500.
  135. PoseTween: Pose-driven Tween Animation. In ACM UIST.
  136. A Comprehensive Study of Weight Sharing in Graph Networks for 3D Human Pose Estimation. In ECCV.
  137. Attention Mechanism Exploits Temporal Contexts: Real-Time 3D Human Pose Reconstruction. In CVPR.
  138. Deep dual consecutive network for human pose estimation. In CVPR.
  139. Temporal Feature Alignment and Mutual Information Maximization for Video-Based Human Pose Estimation. In CVPR.
  140. A survey of human pose estimation: the body parts parsing based methods. In JVCIR, Vol. 32. 10–19.
  141. Fully Convolutional Networks for Semantic Segmentation. In CVPR.
  142. SMPL: A Skinned Multi-Person Linear Model. In ACM TOG, Vol. 34. 1–16.
  143. Vision-based Estimation of MDS-UPDRS Gait Scores for Assessing Parkinson’s Disease Motor Severity. In arXiv preprint arXiv:2007.08920.
  144. Lstm pose machines. In CVPR.
  145. Rethinking the heatmap regression for bottom-up human pose estimation. In CVPR.
  146. 2d/3d pose estimation and action recognition using multitask deep learning. In CVPR.
  147. Human pose regression by combining indirect part detection and contextual information. In Computers & Graphics, Vol. 85. 15–22.
  148. Transfusion: Cross-view fusion with transformer for 3d human pose estimation. In BMVC.
  149. PPT: token-Pruned Pose Transformer for monocular and multi-view human pose estimation. In ECCV.
  150. Enhancing Human Pose Estimation in Ancient Vase Paintings via Perceptually-grounded Style Transfer Learning. In arXiv preprint arXiv:2012.05616.
  151. AMASS: Archive of Motion Capture as Surface Shapes. In ICCV.
  152. Tfpose: Direct human pose estimation with transformers. In arXiv preprint arXiv:2103.15320.
  153. Poseur: Direct Human Pose Regression with Transformers. In ECCV.
  154. Graph Embedded Pose Clustering for Anomaly Detection. In CVPR.
  155. A simple yet effective baseline for 3d human pose estimation. In ICCV.
  156. Monocular 3D Human Pose Estimation in the Wild Using Improved CNN Supervision. In 3DV.
  157. XNect: Real-time Multi-Person 3D Motion Capture with a Single RGB Camera. In ACM TOG, Vol. 39. 82–1.
  158. Single-Shot Multi-Person 3D Pose Estimation From Monocular RGB. In 3DV.
  159. Vnect: Real-time 3d human pose estimation with a single rgb camera. In ACM TOG, Vol. 36. 1–14.
  160. Real-time upper body detection and 3D pose estimation in monoscopic images. In ECCV.
  161. Multiview-Consistent Semi-Supervised Learning for 3D Human Pose Estimation. In CVPR.
  162. Thomas B Moeslund and Erik Granum. 2001. A survey of computer vision-based human motion capture. In CVIU, Vol. 81. 231–268.
  163. A survey of advances in vision-based human motion capture and analysis. In CVIU, Vol. 104. 90–126.
  164. Camera Distance-aware Top-down Approach for 3D Multi-person Pose Estimation from a Single RGB Image. In ICCV.
  165. Posefix: Model-agnostic general human pose refinement network. In CVPR.
  166. Gyeongsik Moon and Kyoung Mu Lee. 2020. I2L-MeshNet: Image-to-Lixel Prediction Network for Accurate 3D Human Pose and Mesh Estimation from a Single RGB Image. In ECCV.
  167. Francesc Moreno-Noguer. 2017. 3d human pose estimation from a single image via distance matrix regression. In CVPR.
  168. The Progress of Human Pose Estimation: A Survey and Taxonomy of Models Applied in 2D Human Pose Estimation. In IEEE Access, Vol. 8. 133330–133348.
  169. Associative embedding: End-to-end learning for joint detection and grouping. In NeurIPS.
  170. Stacked hourglass networks for human pose estimation. In ECCV.
  171. Numerical coordinate regression with convolutional neural networks. In arXiv preprint arXiv:1801.07372.
  172. Monocular 3D Human Pose Estimation by Predicting Depth on Joints. In ICCV.
  173. Unsupervised Human 3D Pose Representation with Viewpoint and Pose Disentanglement. In ECCV.
  174. Single-Stage Multi-Person Pose Machines. In ICCV.
  175. Neural Body Fitting: Unifying Deep Learning and Model-Based Human Pose and Shape Estimation. In 3DV.
  176. STAR: A Spare Trained Articulated Human Body Regressor. In ECCV.
  177. Paschalis Panteleris and Antonis Argyros. 2021. PE-former: Pose Estimation Transformer. In arXiv preprint arXiv:2112.04981.
  178. Personlab: Person pose estimation and instance segmentation with a bottom-up, part-based, geometric embedding model. In ECCV.
  179. Towards Accurate Multi-Person Pose Estimation in the Wild. In CVPR.
  180. TailorNet: Predicting Clothing in 3D as a Function of Human Pose, Shape and Garment Style. In CVPR.
  181. Expressive Body Capture: 3D Hands, Face, and Body from a Single Image. In CVPR.
  182. Ordinal Depth Supervision for 3D Human Pose Estimation. In CVPR.
  183. Coarse-to-Fine Volumetric Prediction for Single-Image 3D Human Pose. In CVPR.
  184. Harvesting Multiple Views for Marker-Less 3D Human Pose Annotations. In CVPR.
  185. Learning to Estimate 3D Human Pose and Shape from a Single Color Image. In CVPR.
  186. 3D human pose estimation in video with temporal convolutions and semi-supervised training. In CVPR.
  187. Jointly optimize data augmentation and network training: Adversarial data augmentation in human pose estimation. In CVPR.
  188. Flowing convnets for human pose estimation in videos. In ICCV.
  189. Deep convolutional neural networks for efficient pose estimation in gesture videos. In ACCV.
  190. Domes to Drones: Self-Supervised Active Triangulation for 3D Human Pose Reconstruction. In NeurIPS.
  191. Deepcut: Joint subset partition and labeling for multi person pose estimation. In CVPR.
  192. Dyna: A Model of Dynamic Human Shape in Motion. In ACM TOG, Vol. 34. 1–14.
  193. Ronald Poppe. 2007. Vision-based human motion analysis: An overview. In CVIU, Vol. 108. 4–18.
  194. Ammar Qammaz and Antonis A Argyros. 2019. MocapNET: Ensemble of SNN Encoders for 3D Human Pose Estimation in RGB Images. In BMVC.
  195. Pointnet: Deep learning on point sets for 3d classification and segmentation. In CVPR.
  196. Pointnet++: Deep hierarchical feature learning on point sets in a metric space. In NeurIPS.
  197. Cross View Fusion for 3D Human Pose Estimation. In ICCV.
  198. Peeking into occluded joints: A novel framework for crowd pose estimation. In arXiv preprint arXiv:2003.10506.
  199. Pose machines: Articulated pose estimation via inference machines. In ECCV.
  200. Mir Rayat Imtiaz Hossain and James J. Little. 2018. Exploiting temporal information for 3D human pose estimation. In ECCV.
  201. TesseTrack: End-to-End Learnable Multi-Person Articulated 3D Pose Tracking. In CVPR.
  202. Lightweight Multi-View 3D Pose Estimation Through Camera-Disentangled Representation. In CVPR.
  203. Faster r-cnn: Towards real-time object detection with region proposal networks. In NeurIPS.
  204. Unsupervised Geometry-Aware Representation for 3D Human Pose Estimation. In ECCV.
  205. Learning Monocular 3D Human Pose Estimation From Multi-View Images. In CVPR.
  206. LCR-Net: Localization-Classification-Regression for Human Pose. In CVPR.
  207. LCR-Net++: Multi-person 2D and 3D Pose Detection in Natural Images. In IEEE TPAMI, Vol. 42. 1146–1161.
  208. Sebastian Ruder. 2017. An overview of multi-task learning in deep neural networks. In arXiv preprint arXiv:1706.05098.
  209. Markerless Outdoor Human Motion Capture Using Multiple Autonomous Micro Aerial Vehicles. In ICCV.
  210. PIFu: Pixel-Aligned Implicit Function for High-Resolution Clothed Human Digitization. In ICCV.
  211. PIFuHD: Multi-Level Pixel-Aligned Implicit Function for High-Resolution 3D Human Digitization. In CVPR.
  212. Ben Sapp and Ben Taskar. 2013. Modec: Multimodal decomposable models for human pose estimation. In CVPR.
  213. 3d human pose estimation: A review of the literature and analysis of covariates. In CVIU, Vol. 152. 1–20.
  214. Monocular 3D Human Pose Estimation by Generation and Ordinal Ranking. In ICCV.
  215. End-to-End Multi-Person Pose Estimation With Transformers. In CVPR.
  216. HumanEva: Synchronized video and motion capture dataset and baseline algorithm for evaluation of articulated human motion. In IJCV, Vol. 87. 4.
  217. 15 keypoints is all you need. In CVPR.
  218. Multi-person pose estimation with enhanced channel-wise and spatial information. In CVPR.
  219. View-Invariant Probabilistic Embedding for Human Pose. In ECCV.
  220. Deep high-resolution representation learning for human pose estimation. In CVPR.
  221. Compositional human pose regression. In ICCV.
  222. Wei Tang and Ying Wu. 2019. Does Learning Specific Features for Related Parts Help Human Pose Estimation?. In CVPR.
  223. Deeply learned compositional models for human pose estimation. In ECCV.
  224. Structured Prediction of 3D Human Pose with Deep Neural Networks. In BMVC.
  225. Learning to fuse 2d and 3d image cues for monocular body pose estimation. In ICCV.
  226. Direct Prediction of 3D Body Poses From Motion Compensated Sequences. In CVPR.
  227. DirectPose: Direct End-to-End Multi-Person Pose Estimation. In arXiv preprint arXiv:1911.07451.
  228. SelfPose: 3D Egocentric Pose Estimation from a Headset Mounted Camera. In arXiv preprint arXiv:2011.01519.
  229. xR-EgoPose: Egocentric 3D Human Pose from an HMD Camera. In ICCV.
  230. Efficient object localization using convolutional networks. In CVPR.
  231. Joint training of a convolutional network and a graphical model for human pose estimation. In NeurIPS.
  232. Alexander Toshev and Christian Szegedy. 2014. Deeppose: Human pose estimation via deep neural networks. In CVPR.
  233. Total Capture: 3D Human Pose Estimation Fusing Video and Inertial Sensors. In BMVC.
  234. VoxelPose: Towards Multi-Camera 3D Human Pose Estimation in Wild Environment. In ECCV.
  235. Self-Supervised Learning of Motion Capture. In NeurIPS.
  236. Self-supervised Keypoint Correspondences for Multi-Person Pose Estimation and Tracking in Videos. In arXiv preprint arXiv:2004.12652.
  237. Learning from synthetic humans. In CVPR.
  238. Ignas Budvytis Vince Tan and Roberto Cipolla. 2017. Indirect deep structured learning for 3D human body shape and pose prediction. In BMVC.
  239. Recovering Accurate 3D Human Pose in The Wild Using IMUs and a Moving Camera. In ECCV.
  240. Sparse inertial poser: Automatic 3d human pose estimation from sparse imus. In Computer Graphics Forum.
  241. Bastian Wandt and Bodo Rosenhahn. 2019. RepNet: Weakly Supervised Training of an Adversarial Reprojection Network for 3D Human Pose Estimation. In CVPR.
  242. BLSM: A Bone-Level Skinned Model of the Human Mesh. In ECCV.
  243. Regularizing Vector Embedding in Bottom-Up Human Pose Estimation. In ECCV.
  244. Not All Parts Are Created Equal: 3D Pose Estimation by Modeling Bi-Directional Dependencies of Body Parts. In ICCV.
  245. Graph-PCNN: Two Stage Human Pose Estimation with Graph Pose Refinement. In arXiv preprint arXiv:2007.10599.
  246. AI Coach: Deep Human Pose Estimation and Analysis for Personalized Athletic Training Assistance. In ACM MM.
  247. Motion Guided 3D Pose Estimation from Videos. In ECCV.
  248. Sequential 3D Human Pose and Shape Estimation From Point Clouds. In CVPR.
  249. DRPose3D: Depth Ranking in 3D Human Pose Estimation. In IJCAI.
  250. Direct Multi-view Multi-person 3D Human Pose Estimation. In NeurIPS.
  251. Lite pose: Efficient architecture design for 2d human pose estimation. In CVPR.
  252. Distribution-Aware Single-Stage Models for Multi-Person 3D Pose Estimation. In CVPR.
  253. Convolutional pose machines. In CVPR.
  254. Photo Wake-Up: 3D Character Animation From a Single Photo. In CVPR.
  255. Pose2Pose: pose selection and transfer for 2D character animation. In IUI.
  256. Ai challenger: A large-scale dataset for going deeper in image understanding. In arXiv preprint arXiv:1711.06475.
  257. Monocular total capture: Posing face, body, and hands in the wild. In CVPR.
  258. Simple Baselines for Human Pose Estimation and Tracking. In ECCV.
  259. MetaFuse: A Pre-trained Fusion Model for Human Pose Estimation. In CVPR.
  260. A2J: Anchor-to-Joint Regression Network for 3D Articulated Pose Estimation from a Single Depth Image. In ICCV.
  261. GHUM &\&& GHUML: Generative 3D Human Shape and Articulated Pose Models. In CVPR.
  262. Deep Kinematics Analysis for Monocular 3D Human Pose Estimation. In CVPR.
  263. Vipnas: Efficient video pose estimation via neural architecture search. In CVPR.
  264. Mo 2 cap 2: Real-time mobile 3d motion capture with a cap-mounted fisheye camera. In IEEE TVCG Proc. VR, Vol. 25. 2093–2101.
  265. 3D Human Shape and Pose from a Single Low-Resolution Image with Self-Supervised Learning. In ECCV.
  266. Spatial temporal graph convolutional networks for skeleton-based action recognition. In AAAI.
  267. Transpose: Keypoint localization via transformer. In ICCV.
  268. Learning feature pyramids for human pose estimation. In ICCV.
  269. End-to-end learning of deformable mixture of parts and deep convolutional neural networks for human pose estimation. In CVPR.
  270. 3D Human Pose Estimation in the Wild by Adversarial Learning. In CVPR.
  271. Yi Yang and Deva Ramanan. 2012. Articulated human detection with flexible mixtures of parts. In IEEE TPAMI, Vol. 35. 2878–2890.
  272. Faster VoxelPose: Real-time 3D Human Pose Estimation by Orthographic Projection. In ECCV.
  273. Lite-hrnet: A lightweight high-resolution network. In CVPR.
  274. DoubleFusion: Real-time Capture of Human Performances with Inner Body Shapes from a Single Depth Sensor. In IEEE TPAMI. 7287–7296.
  275. Simulcap: Single-view human performance capture with cloth simulation. In CVPR.
  276. Hrformer: High-resolution vision transformer for dense predict. In NeurIPS.
  277. Weakly Supervised 3D Human Pose and Shape Reconstruction with Normalizing Flows. In ECCV.
  278. Monocular 3D Pose and Shape Estimation of Multiple People in Natural Scenes: The Importance of Multiple Scene Constraints. In CVPR.
  279. Deep Network for the Integrated 3D Sensing of Multiple People in Natural Images. In NeurIPS.
  280. SRNet: Improving Generalization in 3D Human Pose Estimation with a Split-and-Recombine Approach. In ECCV.
  281. 3D Human Mesh Regression With Dense Correspondence. In CVPR.
  282. Distribution-aware coordinate representation for human pose estimation. In CVPR.
  283. Fast human pose estimation. In CVPR.
  284. Human pose estimation with spatial contextual information. In arXiv preprint arXiv:1901.01760.
  285. Vid2Player: Controllable Video Sprites that Behave and Appear like Professional Tennis Players. In arXiv preprint arXiv:2008.04524.
  286. PyMAF: 3D Human Pose and Shape Regression with Pyramidal Mesh Alignment Feedback Loop. In ICCV.
  287. MixSTE: Seq2seq Mixed Spatio-Temporal Encoder for 3D Human Pose Estimation in Video. In CVPR.
  288. Object-Occluded Human Shape and Pose Estimation From a Single Color Image. In CVPR.
  289. EfficientPose: Efficient Human Pose Estimation with Neural Architecture Search. In arXiv preprint arXiv:2012.07086.
  290. From actemes to action: A strongly-supervised representation for detailed action understanding. In ICCV.
  291. 4D Association Graph for Realtime Multi-person Motion Capture Using Multiple Video Cameras. In CVPR.
  292. Key Frame Proposal Network for Efficient Pose Estimation in Videos. In arXiv preprint arXiv:2007.15217.
  293. Fusing Wearable IMUs With Multi-View Images for Human Pose Estimation: A Geometric Approach. In CVPR.
  294. AdaFuse: Adaptive Multiview Fusion for Accurate Human Pose Estimation in the Wild. In IJCV, Vol. 129. 703–718.
  295. Semantic Graph Convolutional Networks for 3D Human Pose Regression. In CVPR.
  296. Through-Wall Human Mesh Recovery Using Radio Signals. In ICCV.
  297. RF-based 3D skeletons. In SIGCOMM.
  298. PoseFormerV2: Exploring Frequency Domain for Efficient and Robust 3D Human Pose Estimation. In CVPR.
  299. SMAP: Single-Shot Multi-Person Absolute 3D Pose Estimation. In ECCV.
  300. POTTER: Pooling Attention Transformer for Efficient Human Mesh Recovery. In CVPR.
  301. A Lightweight Graph Transformer Network for Human Mesh Reconstruction from 2D Human Pose. In ACM Multimedia.
  302. FeatER: An Efficient Network for Human Reconstruction via Feature Map-Based TransformER. In CVPR.
  303. 3D Human Pose Estimation with Spatial and Temporal Transformers. In ICCV.
  304. TexMesh: Reconstructing Detailed Human Texture and Geometry from RGB-D Video. In ECCV.
  305. Unsupervised Shape and Pose Disentanglement for 3D Meshes. In arXiv preprint arXiv:2007.11341.
  306. HEMlets Pose: Learning Part-Centric Heatmap Triplets for Accurate 3D Human Pose Estimation. In ICCV.
  307. Towards 3D Human Pose Estimation in the Wild: A Weakly-Supervised Approach. In ICCV.
  308. Deep kinematic pose regression. In ECCV.
  309. Objects as points. In arXiv preprint arXiv:1904.07850.
  310. Sparseness Meets Deepness: 3D Human Pose Estimation from Monocular Video. In CVPR.
  311. Monocap: Monocular human motion capture using a cnn coupled with a geometric prior. In IEEE TPAMI, Vol. 41. 901–914.
  312. Detailed human shape estimation from a single image by hierarchical mesh deformation. In CVPR.
  313. Reconstructing NBA players. In ECCV.
  314. Multi-person pose estimation for posetrack with enhanced part affinity fields. In ICCV PoseTrack Workshop.
  315. Zhiming Zou and Wei Tang. 2021. Modulated Graph Convolutional Network for 3D Human Pose Estimation. In ICCV.
  316. Silvia Zuffi and Michael J. Black. 2015. The Stitched Puppet: A Graphical Model of 3D Human Shape and Pose. In CVPR.
User Edit Pencil Streamline Icon: https://streamlinehq.com
Authors (8)
  1. Ce Zheng (45 papers)
  2. Wenhan Wu (9 papers)
  3. Chen Chen (753 papers)
  4. Taojiannan Yang (26 papers)
  5. Sijie Zhu (27 papers)
  6. Ju Shen (9 papers)
  7. Nasser Kehtarnavaz (15 papers)
  8. Mubarak Shah (208 papers)
Citations (486)
View on Semantic Scholar

Summary

Overview of Deep Learning-Based Human Pose Estimation: A Survey

The survey paper titled "Deep Learning-Based Human Pose Estimation: A Survey" provides a comprehensive analysis of the advancements in human pose estimation (HPE) through deep learning methodologies. The research targets both 2D and 3D pose estimation problems, addressing challenges such as occlusion, depth ambiguities, and insufficient training data. Covering over 260 papers since 2014, this survey is invaluable for researchers aiming to understand the progressive landscape of HPE solutions.

Key Contributions

This survey delineates the landscape of HPE by categorizing approaches based on data sources such as monocular images, videos, and other sensors like depth and inertial measurement units. Important datasets and evaluation metrics are discussed, providing a critical comparison of methods across standard benchmarks. The paper also explores applications in various fields, including augmented reality and healthcare, highlighting both the achievements and persistent challenges in HPE.

Methodological Insights

2D Pose Estimation

The paper categorizes 2D pose estimation into single-person and multi-person scenarios:

  • Single-Person Pose Estimation: It discusses regression-based methods that map images directly to joint coordinates, as well as heatmap-based methods which predict joint probability maps. Heatmap-based approaches, such as the HRNet, have shown impressive performance due to their ability to preserve spatial locality.
  • Multi-Person Pose Estimation: This is divided further into top-down methods, which perform human detection followed by pose estimation, and bottom-up methods, which locate body parts and assemble them into full poses. Top-down techniques usually exhibit higher accuracy, whereas bottom-up techniques offer computational advantages, especially in crowded scenes.

3D Pose Estimation

3D pose estimation is described under single-view and multi-view categorizations:

  • Single-View: Methods are further split into direct approaches that estimate 3D poses directly from images and 2D-to-3D methods that first compute 2D poses. The latter, leveraging strong 2D detectors and lifting transformations, consistently provide better results.
  • Multi-View and Additional Sensors: For overcoming occlusion problems, multi-view setups are addressed, while other sensors like depth cameras and IMUs offer alternative means for constructing robust 3D pose frameworks.

Quantitative and Comparative Analysis

The survey includes an extensive quantitative analysis, comparing the performance of various HPE methods using benchmarks like MPII, COCO, and Human3.6M datasets. Notably, transformer architectures have been explored recently for their ability to capture complex dependencies, achieving competitive results while enhancing computational efficiency.

Implications and Future Directions

The research presents significant implications for real-world applications, including action recognition, healthcare, and virtual reality. Nevertheless, challenges persist, such as improving occlusion handling and achieving efficient solutions deployable in resource-constrained environments. Future directions include domain adaptation, leveraging multi-objective neural architecture search, and enhancing robustness against adversarial attacks.

Conclusion

This survey offers an extensive review of the state-of-the-art in deep learning-based human pose estimation, providing a valuable resource for researchers and practitioners alike. Through its systemic classification and detailed analysis, it lays a foundation for future advances in HPE, encouraging further exploration into diverse and complex application scenarios.

File Document Download Save Streamline Icon: https://streamlinehq.com PDF File Document Download Save Streamline Icon: https://streamlinehq.com Markdown
Github Logo Streamline Icon: https://streamlinehq.com

GitHub

  1. GitHub - zczcwh/DL-HPE (503 stars)