mlsanityData to 7 Oct 2026

FAST3D: Flow-Aware Self-Training for 3D Object Detectors

arXiv 2110.09355 ↗First posted 18 Oct 2021cs.CV cs.ROcited by 10

Not on the map. Why?

Abstract (arXiv)

In the field of autonomous driving, self-training is widely applied to mitigate distribution shifts in LiDAR-based 3D object detectors. This eliminates the need for expensive, high-quality labels whenever the environment changes (e.g., geographic location, sensor setup, weather condition). State-of-the-art self-training approaches, however, mostly ignore the temporal nature of autonomous driving data. To address this issue, we propose a flow-aware self-training method that enables unsupervised domain adaptation for 3D object detectors on continuous LiDAR point clouds. In order to get reliable pseudo-labels, we leverage scene flow to propagate detections through time. In particular, we introduce a flow-based multi-target tracker, that exploits flow consistency to filter and refine resulting tracks. The emerged precise pseudo-labels then serve as a basis for model re-training. Starting with a pre-trained KITTI model, we conduct experiments on the challenging Waymo Open Dataset to demonstrate the effectiveness of our approach. Without any prior target domain knowledge, our results show a significant improvement over the state-of-the-art.

Key facts. Cited by 10 papers in the map (data through 7 Oct 2026). Builds on 54 of its 65 references that are in the map. First posted on arXiv 18 Oct 2021.

Builds on

54 of 65 references.

References · 54

  1. 2014Adam: A Method for Stochastic Optimizationcited by 36,960
  2. 2016PointNet: Deep Learning on Point Sets for 3D Classification and Segmentationcited by 5,223
  3. 2019nuScenes: A multimodal dataset for autonomous drivingcited by 4,485
  4. 2017PointNet++: Deep Hierarchical Feature Learning on Point Sets in a Metric Spacecited by 4,384
  5. 2019Scalability in Perception for Autonomous Driving: Waymo Open Datasetcited by 2,436
  6. 2018PointPillars: Fast Encoders for Object Detection from Point Cloudscited by 1,707
  7. 2017VoxelNet: End-to-End Learning for Point Cloud Based 3D Object Detectioncited by 1,323
  8. 2018PointRCNN: 3D Object Proposal Generation and Detection from Point Cloudcited by 939
  9. 2019PV-RCNN: Point-Voxel Feature Set Abstraction for 3D Object Detectioncited by 890
  10. 2015A Large Dataset to Train Convolutional Networks for Disparity, Optical Flow, and Scene Flow Estimationcited by 878
  11. 2016Multi-View 3D Object Detection Network for Autonomous Drivingcited by 721
  12. 2018Domain Adaptive Faster R-CNN for Object Detection in the Wildcited by 442
38 more
  1. 20203DSSD: Point-based 3D Single Stage Object Detectorcited by 440
  2. 2020Voxel R-CNN: Towards High Performance Voxel-based 3D Object Detectioncited by 409
  3. 2019From Points to Parts: 3D Object Detection from Point Cloud with Part-aware and Part-aggregation Networkcited by 366
  4. 2019PIXOR: Real-time 3D Object Detection from Point Cloudscited by 358
  5. 2020Point-GNN: Graph Neural Network for 3D Object Detection in a Point Cloudcited by 335
  6. 2017Joint 3D Proposal Generation and Object Detection from View Aggregationcited by 325
  7. 2019STD: Sparse-to-Dense 3D Object Detector for Point Cloudcited by 282
  8. 2018Strong-Weak Distribution Alignment for Adaptive Object Detectioncited by 269
  9. 2020Deep Continuous Fusion for Multi-Sensor 3D Object Detectioncited by 263
  10. 2018FlowNet3D: Learning Scene Flow in 3D Point Cloudscited by 241
  11. 2020Multi-Task Multi-Sensor Fusion for 3D Object Detectioncited by 203
  12. 2019End-to-End Multi-View Fusion for 3D Object Detection in LiDAR Point Cloudscited by 171
  13. 2017PointFusion: Deep Sensor Fusion for 3D Bounding Box Estimationcited by 170
  14. 2021ST3D: Self-training for Unsupervised Domain Adaptation on 3D Object Detectioncited by 158
  15. 2020Train in Germany, Test in The USA: Making 3D Object Detectors Generalizecited by 150
  16. 2020Pillar-based Object Detection for Autonomous Drivingcited by 137
  17. 2019Exploring Object Relation in Mean Teacher for Cross-Domain Detectioncited by 136
  18. 2019Progressive Domain Adaptation for Object Detectioncited by 134
  19. 2020Exploring Categorical Regularization for Domain Adaptive Object Detectioncited by 130
  20. 2019xMUDA: Cross-Modal Unsupervised Domain Adaptation for 3D Semantic Segmentationcited by 129
  21. 2019Diversify and Match: A Domain Adaptive Representation Learning Paradigm for Object Detectioncited by 122
  22. 2019Fast Point R-CNNcited by 119
  23. 2019MeteorNet: Deep Learning on Dynamic 3D Point Cloud Sequencescited by 117
  24. 2020Complete & Label: A Domain Adaptation Approach to Semantic Segmentation of LiDAR Point Cloudscited by 116
  25. 2019PointDAN: A Multi-Scale 3D Domain Adaption Network for Point Cloud Representationcited by 114
  26. 2019Multi-adversarial Faster-RCNN for Unrestricted Object Detectioncited by 111
  27. 2019A Robust Learning Approach to Domain Adaptive Object Detectioncited by 109
  28. 2019HPLFlowNet: Hierarchical Permutohedral Lattice FlowNet for Scene Flow Estimation on Large-scale Point Cloudscited by 108
  29. 2020Self-Supervised Learning for Domain Adaptation on Point-Cloudscited by 99
  30. 2019Just Go with the Flow: Self-Supervised Scene Flow Estimationcited by 98
  31. 2019Towards Universal Object Detection by Domain Attentioncited by 89
  32. 2020HVNet: Hybrid Voxel Network for LiDAR Based 3D Object Detectioncited by 82
  33. 2020Weakly Supervised 3D Object Detection from Lidar Point Cloudcited by 67
  34. 2019Automatic adaptation of object detectors to new domains using self-trainingcited by 66
  35. 20163D Fully Convolutional Network for Vehicle Detection in Point Cloudcited by 57
  36. 2021Scalable Scene Flow from Point Clouds in the Real Worldcited by 49
  37. 2019StarNet: Targeted Computation for Object Detection in Point Cloudscited by 46
  38. 2019Domain Adaptation for Object Detection via Style Consistencycited by 45

50 most cited of 54 All 54 (JSON)

Built on by

Cited by year

Citing papers per year: 2021: 1; 2022: 3; 2023: 0; 2024: 4; 2025: 2.

10 papers.

  1. 20223D Object Detection for Autonomous Driving: A Comprehensive Surveycited by 132
  2. 2021The Norm Must Go On: Dynamic Unsupervised Domain Adaptation by Normalizationcited by 131
  3. 2024Backpropagation-free Network for 3D Test-time Adaptationcited by 14
  4. 2022Viewer-Centred Surface Completion for Unsupervised Domain Adaptation in 3D Object Detectioncited by 10
  5. 2024SOAP: Cross-sensor Domain Adaptation for 3D Object Detection Using Stationary Object Aggregation Pseudo-labellingcited by 8
  6. 2024TTT-KD: Test-Time Training for 3D Semantic Segmentation through Knowledge Distillation from Foundation Modelscited by 5
  7. 2022SAILOR: Scaling Anchors via Insights into Latent Object Representationcited by 4
  8. 2024Let-It-Flow: Simultaneous Optimization of 3D Flow and Object Clusteringcited by 3
  9. 2025GBlobs: Explicit Local Structure via Gaussian Blobs for Improved Cross-Domain LiDAR-based 3D Object Detectioncited by 2
  10. 2025Semi-Supervised Diversity-Aware Domain Adaptation for 3D Object detectioncited by 0