mlsanityData to 7 Oct 2026

SelFlow: Self-Supervised Learning of Optical Flow

arXiv 1904.09117 ↗First posted 19 Apr 2019cs.CV cs.LGcited by 128

Not on the map. Why?

Abstract (arXiv)

We present a self-supervised learning approach for optical flow. Our method distills reliable flow estimations from non-occluded pixels, and uses these predictions as ground truth to learn optical flow for hallucinated occlusions. We further design a simple CNN to utilize temporal information from multiple frames for better flow estimation. These two principles lead to an approach that yields the best performance for unsupervised optical flow learning on the challenging benchmarks including MPI Sintel, KITTI 2012 and 2015. More notably, our self-supervised pre-trained model provides an excellent initialization for supervised fine-tuning. Our fine-tuned models achieve state-of-the-art results on all three datasets. At the time of writing, we achieve EPE=4.26 on the Sintel benchmark, outperforming all submitted methods.

Key facts. Cited by 128 papers in the map (data through 7 Oct 2026). First posted on arXiv 19 Apr 2019.

Builds on

No reference list yet.

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2021: 36; 2022: 39; 2023: 21; 2024: 20; 2025: 7; 2026: 5.

128 papers.

Sort:

By area:

  1. 2021Learning to Estimate Hidden Motions with Global Motion Aggregationcited by 223
  2. 2022CroCo v2: Improved Cross-view Completion Pre-training for Stereo Matching and Optical Flowcited by 165
  3. 2021E-RAFT: Dense Optical Flow from Event Camerascited by 149
  4. 2021Self-supervised Video Object Segmentation by Motion Groupingcited by 147
  5. 2024CoTracker3: Simpler and Better Point Tracking by Pseudo-Labelling Real Videoscited by 122
  6. 2024BootsTAP: Bootstrapped Training for Tracking-Any-Pointcited by 115
  7. 2022IFRNet: Intermediate Feature Refine Network for Efficient Frame Interpolationcited by 107
  8. 2021AutoFlow: Learning a Better Training Set for Optical Flowcited by 102
  9. 2022CRAFT: Cross-Attentional Flow Transformer for Robust Optical Flowcited by 82
  10. 2023VideoFlow: Exploiting Temporal Cues for Multi-frame Optical Flow Estimationcited by 82
  11. 2021PDC-Net+: Enhanced Probabilistic Dense Correspondence Networkcited by 80
  12. 2021SMURF: Self-Teaching Multi-Frame Unsupervised RAFT with Full-Image Warpingcited by 74
38 more
  1. 2022Learning Dense and Continuous Optical Flow from an Event Cameracited by 59
  2. 2021Learning Monocular Depth in Dynamic Scenes via Instance-Aware Projection Consistencycited by 58
  3. 2021Multiple Instance Captioning: Learning Representations from Histopathology Textbooks and Articlescited by 58
  4. 2024MemFlow: Optical Flow Estimation and Prediction with Memorycited by 55
  5. 2023MC-JEPA: A Joint-Embedding Predictive Architecture for Self-Supervised Learning of Motion and Content Featurescited by 53
  6. 2023Synthetic Datasets for Autonomous Driving: A Surveycited by 48
  7. 2022SSMTL++: Revisiting Self-Supervised Multi-Task Learning for Video Anomaly Detectioncited by 45
  8. 2021Warp Consistency for Unsupervised Learning of Dense Correspondencescited by 42
  9. 2021Self-Point-Flow: Self-Supervised Scene Flow Estimation from Point Clouds with Optimal Transport and Random Walkcited by 39
  10. 2022RNNPose: Recurrent 6-DoF Object Pose Refinement with Robust Correspondence Field Estimation and Pose Optimizationcited by 38
  11. 2022Bi-PointFlowNet: Bidirectional Learning for Point Cloud Based Scene Flow Estimationcited by 38
  12. 2022Learning Pixel Trajectories with Multiscale Contrastive Random Walkscited by 36
  13. 2023SkyEye: Self-Supervised Bird's-Eye-View Semantic Mapping Using Monocular Frontal View Imagescited by 35
  14. 2022Disentangling Architecture and Training for Optical Flowcited by 32
  15. 2022Learning Optical Flow, Depth, and Scene Flow without Real-World Labelscited by 31
  16. 2024Cross-View Completion Models are Zero-shot Correspondence Estimatorscited by 30
  17. 2021Self-Supervised Multi-Frame Monocular Scene Flowcited by 29
  18. 2021Optical Flow Estimation for Spiking Cameracited by 29
  19. 2021Learning optical flow from still imagescited by 28
  20. 2022OcclusionFusion: Occlusion-aware Motion Estimation for Real-time Dynamic 3D Reconstructioncited by 27
  21. 2022DiffPoseNet: Direct Differentiable Camera Pose Estimationcited by 27
  22. 2021DeepTag: An Unsupervised Deep Learning Method for Motion Tracking on Cardiac Tagging Magnetic Resonance Imagescited by 26
  23. 2021Self-supervised Geometric Perceptioncited by 26
  24. 2022PanoFlow: Learning 360deg Optical Flow for Surrounding Temporal Understandingcited by 26
  25. 2022DyAnNet: A Scene Dynamicity Guided Self-Trained Video Anomaly Detection Networkcited by 26
  26. 2021Fusion-FlowNet: Energy-Efficient Optical Flow Estimation using Sensor Fusion and Deep Fused Spiking-Analog Network Architecturescited by 25
  27. 2021SVDistNet: Self-Supervised Near-Field Distance Estimation on Surround View Fisheye Camerascited by 25
  28. 2022Imposing Consistency for Optical Flow Estimationcited by 24
  29. 2022RealFlow: EM-based Realistic Optical Flow Dataset Generation from Videoscited by 22
  30. 2024VFIMamba: Video Frame Interpolation with State Space Modelscited by 22
  31. 2021The Functional Correspondence Problemcited by 20
  32. 2023AccFlow: Backward Accumulation for Long-Range Optical Flowcited by 19
  33. 2021VOLDOR: Visual Odometry from Log-logistic Dense Optical flow Residualscited by 17
  34. 2022SelfD: Self-Learning Large-Scale Driving Policies From the Webcited by 16
  35. 2021GyroFlow: Gyroscope-Guided Unsupervised Optical Flow Learningcited by 15
  36. 2022Simulating Fluids in Real-World Still Imagescited by 15
  37. 2022Unsupervised Deformable Image Registration with Absent Correspondences in Pre-operative and Post-Recurrence Brain Tumor MRI Scanscited by 15
  38. 2022A Perceptual Quality Metric for Video Frame Interpolationcited by 15

50 most cited of 128 All 128 (JSON)