mlsanityData to 14 Oct 2026

RAFT-3D: Scene Flow using Rigid-Motion Embeddings

arXiv 2012.00726 ↗First posted 1 Dec 2020cs.CVcited by 122

Not on the map. Why?

Abstract (arXiv)

We address the problem of scene flow: given a pair of stereo or RGB-D video frames, estimate pixelwise 3D motion. We introduce RAFT-3D, a new deep architecture for scene flow. RAFT-3D is based on the RAFT model developed for optical flow but iteratively updates a dense field of pixelwise SE3 motion instead of 2D motion. A key innovation of RAFT-3D is rigid-motion embeddings, which represent a soft grouping of pixels into rigid objects. Integral to rigid-motion embeddings is Dense-SE3, a differentiable layer that enforces geometric consistency of the embeddings. Experiments show that RAFT-3D achieves state-of-the-art performance. On FlyingThings3D, under the two-view evaluation, we improved the best published accuracy (d < 0.05) from 34.3% to 83.7%. On KITTI, we achieve an error of 5.77, outperforming the best published method (6.31), despite using no object instance supervision. Code is available at https://github.com/princeton-vl/RAFT-3D.

Key facts. Cited by 122 papers in the map (data through 14 Oct 2026). First posted on arXiv 1 Dec 2020.

Builds on

No reference list yet.

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2021: 6; 2022: 20; 2023: 27; 2024: 30; 2025: 28; 2026: 11.

122 papers.

Sort:

By area:

  1. 2024Shape of Motion: 4D Reconstruction from a Single Videocited by 250
  2. 2022Deep Patch Visual Odometrycited by 190
  3. 2024SpatialTracker: Tracking Any 2D Pixels in 3D Spacecited by 181
  4. 2023AMT: All-Pairs Multi-Field Transforms for Efficient Frame Interpolationcited by 180
  5. 2023Spring: A High-Resolution High-Detail Dataset and Benchmark for Scene Flow, Optical Flow and Stereocited by 172
  6. 2023Infinite Photorealistic Worlds using Procedural Generationcited by 147
  7. 2024Infinigen Indoors: Photorealistic Indoor Scenes using Procedural Generationcited by 137
  8. 2021Neural Scene Flow Priorcited by 91
  9. 2024Stereo4D: Learning How Things Move in 3D from Internet Stereo Videoscited by 89
  10. 2025St4RTrack: Simultaneous 4D Reconstruction and Tracking in the Worldcited by 89
  11. 2025TAPIP3D: Tracking Any Point in Persistent 3D Geometrycited by 69
  12. 2024DELTA: Dense Efficient Long-range 3D Tracking for any videocited by 65
38 more
  1. 2021CamLiFlow: Bidirectional Camera-LiDAR Fusion for Joint Optical Flow and Scene Flow Estimationcited by 56
  2. 2021IterMVS: Iterative Probability Estimation for Efficient Multi-View Stereocited by 56
  3. 2023IEBins: Iterative Elastic Bins for Monocular Depth Estimationcited by 54
  4. 2023Zero-Shot Robot Manipulation from Passive Human Videoscited by 51
  5. 2022Coupled Iterative Refinement for 6D Multi-Object Pose Estimationcited by 48
  6. 2022Dynamic 3D Scene Analysis by Point Cloud Accumulationcited by 47
  7. 2025Dynamic Point Maps: A Versatile Representation for Dynamic 3D Reconstructioncited by 44
  8. 2022IFOR: Iterative Flow Minimization for Robotic Object Rearrangementcited by 42
  9. 2023Fast Neural Scene Flowcited by 41
  10. 2025GS-DiT: Advancing Video Generation with Pseudo 4D Gaussian Fields through Efficient Dense 3D Point Trackingcited by 39
  11. 2022Bi-PointFlowNet: Bidirectional Learning for Point Cloud Based Scene Flow Estimationcited by 38
  12. 2025Instant Gaussian Stream: Fast and Generalizable Streaming of Dynamic Scene Reconstruction via Gaussian Splattingcited by 36
  13. 2025Any4D: Unified Feed-Forward Metric 4D Reconstructioncited by 35
  14. 2025Trace Anything: Representing Any Video in 4D via Trajectory Fieldscited by 33
  15. 2022Disentangling Architecture and Training for Optical Flowcited by 32
  16. 2022Learning Optical Flow, Depth, and Scene Flow without Real-World Labelscited by 31
  17. 2025Pre-training Auto-regressive Robotic Models with 4D Representationscited by 27
  18. 2021Temporally Consistent Online Depth Estimation in Dynamic Scenescited by 26
  19. 2023RPEFlow: Multimodal Fusion of RGB-PointCloud-Event for Joint Optical Flow and Scene Flow Estimationcited by 26
  20. 2024SceneTracker: Long-term Scene Flow Estimation Networkcited by 26
  21. 2023Self-Supervised 3D Scene Flow Estimation Guided by Superpointscited by 24
  22. 2024Match-Stereo-Videos: Bidirectional Alignment for Consistent Dynamic Stereo Matchingcited by 24
  23. 2021SCTN: Sparse Convolution-Transformer Network for Scene Flow Estimationcited by 23
  24. 2024A Survey on Deep Stereo Matching in the Twentiescited by 23
  25. 2025Zero-Shot Monocular Scene Flow Estimation in the Wildcited by 22
  26. 2024Learning-based Multi-View Stereo: A Surveycited by 18
  27. 2025AnyCalib: On-Manifold Learning for Model-Agnostic Single-View Camera Calibrationcited by 18
  28. 2022Depth Field Networks for Generalizable Multi-view Scene Representationcited by 17
  29. 2023Re-Evaluating LiDAR Scene Flow for Autonomous Drivingcited by 17
  30. 2025FlowDreamer: A RGB-D World Model with Flow-based Motion Representations for Robot Manipulationcited by 17
  31. 2025Multi-View 3D Point Trackingcited by 17
  32. 2022Consistent Direct Time-of-Flight Video Depth Super-Resolutioncited by 16
  33. 2023Neural Spline Fields for Burst Image Fusion and Layer Separationcited by 16
  34. 2024FastMAC: Stochastic Spectral Sampling of Correspondence Graphcited by 16
  35. 2023Deep Graph-based Spatial Consistency for Robust Non-rigid Point Cloud Registrationcited by 15
  36. 2024Bring Event into RGB and LiDAR: Hierarchical Visual-Motion Fusion for Scene Flowcited by 15
  37. 2025DGS-LRM: Real-Time Deformable 3D Gaussian Reconstruction From Monocular Videoscited by 15
  38. 2025ReSplat: Learning Recurrent Gaussian Splattingcited by 15

50 most cited of 122 All 122 (JSON)