mlsanityData to 14 Oct 2026

A Large Dataset to Train Convolutional Networks for Disparity, Optical Flow, and Scene Flow Estimation

arXiv 1512.02134 ↗First posted 7 Dec 2015cs.CV cs.LG stat.MLcited by 881

Not on the map. Why?

Cited by 28 in Diffusion models, but placed by its own abstract, which didn’t fit.

Abstract (arXiv)

Recent work has shown that optical flow estimation can be formulated as a supervised learning task and can be successfully solved with convolutional networks. Training of the so-called FlowNet was enabled by a large synthetically generated dataset. The present paper extends the concept of optical flow estimation via convolutional networks to disparity and scene flow estimation. To this end, we propose three synthetic stereo video datasets with sufficient realism, variation, and size to successfully train large networks. Our datasets are the first large-scale datasets to enable training and evaluating scene flow methods. Besides the datasets, we present a convolutional network for real-time disparity estimation that provides state-of-the-art results. By combining a flow and disparity estimation network and training it jointly, we demonstrate the first scene flow estimation with a convolutional network.

Key facts. Cited by 881 papers in the map (data through 14 Oct 2026). First posted on arXiv 7 Dec 2015.

Builds on

No reference list yet.

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2021: 160; 2022: 169; 2023: 147; 2024: 177; 2025: 141; 2026: 87.

881 papers.

Sort:

By area:

  1. 2023DUSt3R: Geometric 3D Vision Made Easycited by 1,567
  2. 2024Grounding Image Matching in 3D with MASt3Rcited by 983
  3. 2023CoTracker: It is Better to Track Togethercited by 522
  4. 2022Kubric: A scalable dataset generatorcited by 381
  5. 2021RAFT-Stereo: Multilevel Recurrent Field Transforms for Stereo Matchingcited by 353
  6. 2021GMFlow: Learning Optical Flow via Global Matchingcited by 327
  7. 2023Human Preference Score: Better Aligning Text-to-Image Models with Human Preferencecited by 287
  8. 2023TAPIR: Tracking Any Point with per-frame Initialization and temporal Refinementcited by 287
  9. 20243D Reconstruction with Spatial Memorycited by 277
  10. 2023PointOdyssey: A Large-Scale Synthetic Dataset for Long-Term Point Trackingcited by 276
  11. 2022FlowFormer: A Transformer Architecture for Optical Flowcited by 275
  12. 2022Unifying Flow, Stereo and Depth Estimationcited by 275
38 more
  1. 2022TAP-Vid: A Benchmark for Tracking Any Point in a Videocited by 263
  2. 2025SAM 3D: 3Dfy Anything in Imagescited by 247
  3. 2021Learning to Estimate Hidden Motions with Global Motion Aggregationcited by 224
  4. 2022Particle Video Revisited: Tracking Through Occlusions Using Point Trajectoriescited by 217
  5. 2022Practical Stereo Matching via Cascaded Recurrent Network with Adaptive Correlationcited by 197
  6. 2021The Temporal Opportunist: Self-Supervised Multi-Frame Monocular Depthcited by 190
  7. 2023Iterative Geometry Encoding Volume for Stereo Matchingcited by 190
  8. 2024SpatialTracker: Tracking Any 2D Pixels in 3D Spacecited by 181
  9. 2023StableRep: Synthetic Images from Text-to-Image Models Make Strong Visual Representation Learnerscited by 177
  10. 2024SEA-RAFT: Simple, Efficient, Accurate RAFT for Optical Flowcited by 174
  11. 2025FoundationStereo: Zero-Shot Stereo Matchingcited by 173
  12. 2023Spring: A High-Resolution High-Detail Dataset and Benchmark for Scene Flow, Optical Flow and Stereocited by 172
  13. 2022CroCo v2: Improved Cross-view Completion Pre-training for Stereo Matching and Optical Flowcited by 166
  14. 2024Tora: Trajectory-oriented Diffusion Transformer for Video Generationcited by 164
  15. 2022CroCo: Self-Supervised Pre-training for 3D Vision Tasks by Cross-View Completioncited by 160
  16. 2021CFNet: Cascade and Fused Cost Volume for Robust Stereo Matchingcited by 154
  17. 2023DynamicStereo: Consistent Dynamic Depth from Stereo Videoscited by 148
  18. 2023Infinite Photorealistic Worlds using Procedural Generationcited by 147
  19. 2022SHIFT: A Synthetic Driving Dataset for Continuous Multi-Task Domain Adaptationcited by 146
  20. 2025MUSt3R: Multi-view Network for Stereo 3D Reconstructioncited by 141
  21. 2021A Survey on Deep Learning Technique for Video Segmentationcited by 132
  22. 20223D Object Detection for Autonomous Driving: A Comprehensive Surveycited by 132
  23. 2022Attention Concatenation Volume for Accurate and Efficient Stereo Matchingcited by 130
  24. 2021Modular Interactive Video Object Segmentation: Interaction-to-Mask, Propagation and Difference-Aware Fusioncited by 129
  25. 2022QuadTree Attention for Vision Transformerscited by 129
  26. 20214DComplete: Non-Rigid Motion Estimation Beyond the Observable Surfacecited by 124
  27. 2021SMD-Nets: Stereo Mixture Density Networkscited by 120
  28. 2023The Surprising Effectiveness of Diffusion Models for Optical Flow and Monocular Depth Estimationcited by 116
  29. 2023FlowFormer++: Masked Cost Volume Autoencoding for Pretraining Optical Flow Estimationcited by 114
  30. 2021Robust Learning Meets Generative Models: Can Proxy Distributions Improve Adversarial Robustness?cited by 113
  31. 2022P3Depth: Monocular Depth Estimation with a Piecewise Planarity Priorcited by 107
  32. 2021Generative Models as a Data Source for Multiview Representation Learningcited by 104
  33. 2021Stereo Correspondence and Reconstruction of Endoscopic Data Challengecited by 103
  34. 2021AutoFlow: Learning a Better Training Set for Optical Flowcited by 102
  35. 2024Selective-Stereo: Adaptive Frequency Information Selection for Stereo Matchingcited by 99
  36. 2021Learning to See by Looking at Noisecited by 92
  37. 2022ParticleSfM: Exploiting Dense Point Trajectories for Localizing Moving Cameras in the Wildcited by 92
  38. 2021Neural Scene Flow Priorcited by 91

50 most cited of 881 All 881 (JSON)