new_trends.md

New Trends

• Label Efficient
  • Unsupervised / Self-supervised learning: CO3 [1]
  • Weakly supervised learning: SSC3OD [2]
  • Domain adaption: S2R-ViT [3], DUSA [4]
• Model Adaptation
  • MACP [5]
• Open Heterogeneous Collaborative Perception
  • HEAL [6]
• New Perception Tasks
  • Multi-Object Cooperative Tracking: DMSTrack [7], MOT-CUP [8]
  • Collaborative Semantic Occupancy Prediction: CoHFF [9]

References

1. CO3: Cooperative Unsupervised 3D Representation Learning for Autonomous Driving (ICLR'23) [pdf] [code]
2. SSC3OD: Sparsely Supervised Collaborative 3D Object Detection from LiDAR Point Clouds (SMC'23) [pdf]
3. S2R-ViT for Multi-Agent Cooperative Perception: Bridging the Gap from Simulation to Reality (arXiv'23) [pdf]
4. DUSA: Decoupled Unsupervised Sim2Real Adaptation for Vehicle-to-Everything Collaborative Perception (MM'23) [pdf] [code]
5. MACP: Efficient Model Adaptation for Cooperative Perception (WACV'24) [pdf] [code]
6. HEAL: An Extensible Framework for Open Heterogeneous Collaborative Perception (ICLR'24) [pdf] [code]
7. Probabilistic 3D Multi-Object Cooperative Tracking for Autonomous Driving via Differentiable Multi-Sensor Kalman Filter (ICRA'24) [pdf] [code]
8. Collaborative Multi-Object Tracking with Conformal Uncertainty Propagation (RAL'24) [pdf] [code]
9. Collaborative Semantic Occupancy Prediction with Hybrid Feature Fusion in Connected Automated Vehicles (CVPR'24) [pdf] [code]