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

Semi-Supervised Class-Agnostic Motion Prediction with Pseudo Label Regeneration and BEVMix (2312.08009v2)

Published 13 Dec 2023 in cs.CV

Abstract: Class-agnostic motion prediction methods aim to comprehend motion within open-world scenarios, holding significance for autonomous driving systems. However, training a high-performance model in a fully-supervised manner always requires substantial amounts of manually annotated data, which can be both expensive and time-consuming to obtain. To address this challenge, our study explores the potential of semi-supervised learning (SSL) for class-agnostic motion prediction. Our SSL framework adopts a consistency-based self-training paradigm, enabling the model to learn from unlabeled data by generating pseudo labels through test-time inference. To improve the quality of pseudo labels, we propose a novel motion selection and re-generation module. This module effectively selects reliable pseudo labels and re-generates unreliable ones. Furthermore, we propose two data augmentation strategies: temporal sampling and BEVMix. These strategies facilitate consistency regularization in SSL. Experiments conducted on nuScenes demonstrate that our SSL method can surpass the self-supervised approach by a large margin by utilizing only a tiny fraction of labeled data. Furthermore, our method exhibits comparable performance to weakly and some fully supervised methods. These results highlight the ability of our method to strike a favorable balance between annotation costs and performance. Code will be available at https://github.com/kwwcv/SSMP.

Definition Search Book Streamline Icon: https://streamlinehq.com
References (42)
  1. Pseudo-labeling and confirmation bias in deep semi-supervised learning. In International Joint Conference on Neural Networks (IJCNN), 1–8. IEEE.
  2. Learning with pseudo-ensembles. In Neural Information Processing Systems (NeurIPS). Curran Associates, Inc.
  3. Pointflownet: Learning representations for rigid motion estimation from point clouds. In IEEE Conference on Computer Vision and Pattern Recognition (CVPR). IEEE.
  4. Remixmatch: Semi-supervised learning with distribution alignment and augmentation anchoring. In International Conference on Learning Representations (ICLR).
  5. Mixmatch: A holistic approach to semi-supervised learning. In Neural Information Processing Systems (NeurIPS). Curran Associates, Inc.
  6. nuScenes: A Multimodal Dataset for Autonomous Driving. In IEEE Conference on Computer Vision and Pattern Recognition (CVPR). IEEE.
  7. MultiPath: Multiple Probabilistic Anchor Trajectory Hypotheses for Behavior Prediction. arXiv:1910.05449.
  8. Argoverse: 3d tracking and forecasting with rich maps. In IEEE Conference on Computer Vision and Pattern Recognition (CVPR). IEEE.
  9. Pointmixup: Augmentation for point clouds. In European Conference on Computer Vision (ECCV), 330–345. Springer.
  10. Cuturi, M. 2013. Sinkhorn distances: Lightspeed computation of optimal transport. In Neural Information Processing Systems (NeurIPS). Curran Associates, Inc.
  11. Uncertainty-aware short-term motion prediction of traffic actors for autonomous driving. In IEEE Winter Conference on Applications of Computer Vision (WACV). IEEE.
  12. Tpnet: Trajectory proposal network for motion prediction. In IEEE Conference on Computer Vision and Pattern Recognition (CVPR). IEEE.
  13. HPLFlowNet: Hierarchical Permutohedral Lattice FlowNet for Scene Flow Estimation on Large-Scale Point Clouds. In IEEE Conference on Computer Vision and Pattern Recognition (CVPR). IEEE.
  14. Adam: A Method for Stochastic Optimization. arXiv:1412.6980.
  15. Lasermix for semi-supervised lidar semantic segmentation. In IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 21705–21715. IEEE.
  16. Temporal ensembling for semi-supervised learning. In International Conference on Learning Representations (ICLR).
  17. Pointpillars: Fast encoders for object detection from point clouds. In IEEE Conference on Computer Vision and Pattern Recognition (CVPR). IEEE.
  18. Lee, D.-H.; et al. 2013. Pseudo-label: The simple and efficient semi-supervised learning method for deep neural networks. In Workshop on challenges in representation learning, ICML, volume 3.
  19. Patchwork++: Fast and Robust Ground Segmentation Solving Partial Under-Segmentation Using 3D Point Cloud. In IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS).
  20. Weakly Supervised Class-Agnostic Motion Prediction for Autonomous Driving. In IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 17599–17608. IEEE.
  21. Pnpnet: End-to-end perception and prediction with tracking in the loop. In IEEE Conference on Computer Vision and Pattern Recognition (CVPR). IEEE.
  22. FlowNet3D: Learning Scene Flow in 3D Point Clouds. In IEEE Conference on Computer Vision and Pattern Recognition (CVPR). IEEE.
  23. Self-Supervised Pillar Motion Learning for Autonomous Driving. In IEEE Conference on Computer Vision and Pattern Recognition (CVPR). IEEE.
  24. Mix3d: Out-of-context data augmentation for 3d scenes. In 2021 International Conference on 3D Vision (3DV), 116–125. IEEE.
  25. Pytorch: An imperative style, high-performance deep learning library. In Neural Information Processing Systems (NeurIPS). Curran Associates, Inc.
  26. Long-term occupancy grid prediction using recurrent neural networks. In International Conference on Robotics and Automation (ICRA).
  27. Convolutional LSTM Network: A Machine Learning Approach for Precipitation Nowcasting. In Neural Information Processing Systems (NeurIPS). Curran Associates, Inc.
  28. Fixmatch: Simplifying semi-supervised learning with consistency and confidence. In Neural Information Processing Systems (NeurIPS). Curran Associates, Inc.
  29. A Simple Semi-Supervised Learning Framework for Object Detection. arXiv:2005.04757.
  30. Pwc-net: Cnns for optical flow using pyramid, warping, and cost volume. In IEEE Conference on Computer Vision and Pattern Recognition (CVPR). IEEE.
  31. Mean teachers are better role models: Weight-averaged consistency targets improve semi-supervised deep learning results. In Neural Information Processing Systems (NeurIPS). Curran Associates, Inc.
  32. Hierarchical attention learning of scene flow in 3d point clouds. IEEE Transactions on Image Processing (TIP), 30: 5168–5181.
  33. BE-STI: Spatial-Temporal Integrated Network for Class-agnostic Motion Prediction with Bidirectional Enhancement. In IEEE Conference on Computer Vision and Pattern Recognition (CVPR). IEEE.
  34. Spatiotemporal Transformer Attention Network for 3D Voxel Level Joint Segmentation and Motion Prediction in Point Cloud. In IEEE Intelligent Vehicles Symposium (IV). IEEE.
  35. MotionNet: Joint Perception and Motion Prediction for Autonomous Driving Based on Bird’s Eye View Maps. In IEEE Conference on Computer Vision and Pattern Recognition (CVPR). IEEE.
  36. Unsupervised data augmentation for consistency training. In Neural Information Processing Systems (NeurIPS). Curran Associates, Inc.
  37. End-to-end semi-supervised object detection with soft teacher. In IEEE Conference on International Conference on Computer Vision (ICCV). IEEE.
  38. Cutmix: Regularization strategy to train strong classifiers with localizable features. In IEEE Conference on Computer Vision and Pattern Recognition (CVPR). IEEE.
  39. mixup: Beyond Empirical Risk Minimization. arXiv:1710.09412.
  40. Pointcutmix: Regularization strategy for point cloud classification. Neurocomputing, 505: 58–67.
  41. Instant-teaching: An end-to-end semi-supervised object detection framework. In IEEE Conference on Computer Vision and Pattern Recognition (CVPR). IEEE.
  42. VoxelNet: End-to-End Learning for Point Cloud Based 3D Object Detection. In IEEE Conference on Computer Vision and Pattern Recognition (CVPR). IEEE.
User Edit Pencil Streamline Icon: https://streamlinehq.com
Authors (8)
  1. Kewei Wang (15 papers)
  2. Yizheng Wu (36 papers)
  3. Zhiyu Pan (24 papers)
  4. Xingyi Li (14 papers)
  5. Ke Xian (26 papers)
  6. Zhe Wang (574 papers)
  7. Zhiguo Cao (88 papers)
  8. Guosheng Lin (157 papers)
Citations (5)
View on Semantic Scholar

Summary

We haven't generated a summary for this paper yet.

Ai Sparkles Streamline Icon: https://streamlinehq.com Summarize Now