Research map / Video / Video segmentation and tracking
Slow Flow: Exploiting High-Speed Cameras for Accurate and Diverse Optical Flow Reference Data
Joel Janai, Fatma Guney, Jonas Wulff, Michael J. Black, Andreas Geiger
Check your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| use the full dense tracking model rather than Epic Flow at 24fps | Slow Flow · MPI Sintel, all regions | EPE: 5.53 -> 2.58 | Improved |
| use the full dense tracking model rather than Epic Flow at 24fps | Slow Flow · MPI Sintel, occluded regions | EPE: 16.54 -> 9.45 | Improved |
| use the full dense tracking model rather than Epic Flow at 24fps | Slow Flow · MPI Sintel, visible regions | EPE: 2.45 -> 0.87 | Improved |
| add dense tracking with temporal and occlusion reasoning instead of accumulating Flowlets at 144fps (ablation) | Slow Flow · MPI Sintel, all regions | EPE: 4.03 -> 2.58 | Improved |
| add dense tracking with temporal and occlusion reasoning instead of accumulating Flowlets at 144fps (ablation) | Slow Flow · MPI Sintel, occluded regions | EPE: 15.24 -> 9.45 | Improved |
| accumulate multi-frame Flowlets at 1008fps rather than choosing a coarser frame rate (ablation) | Slow Flow · MPI Sintel, all regions | EPE: 4.03 -> 5.38 | Hurt |
| accumulate multi-frame Flowlets at 1008fps rather than 144fps (ablation) | Slow Flow · MPI Sintel, visible regions | EPE: 0.78 -> 1.35 | Hurt |
| accumulate multi-frame Flowlets at 1008fps rather than 144fps (ablation) | Slow Flow · MPI Sintel, motion-boundary edges | EPE: 11.78 -> 2.60 | Improved |
| estimate occlusions with the proposed method rather than a forward/backward EpicFlow check | Slow Flow · MPI Sintel | occlusion Jaccard Index: 62.86% -> 70.09% | Improved |
| use Slow Flow rather than Epic Flow | Optical flow on real-world static scenes · Flow Magnitude 100 | EPE: 1.54 -> 1.47 | No clear effect |
| use Slow Flow rather than Epic Flow | Optical flow on real-world static scenes · Flow Magnitude 300 | EPE: 25.11 -> 5.13 | Improved |
| use realistic motion-blurred inputs | matching-based optical flow methods · Real-world benchmark, varying motion magnitudes | EPE | Hurt |
Later papers that compared with this method
1 later accepted paper reports SlowFlow as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.
- AccFlow: Backward Accumulation for Long-Range Optical Flow (ICCV 2023 · better on 1 of 1 shared result)
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- On the real-world generalisability of Optical Flow models (ECCV 2026 · similarity 0.90)
- Learning Large Motion Estimation from Intermediate Representations with a High-Resolution Optical Flow Dataset Featuring Long-Range Dynamic Motion (ICCV 2025 · similarity 0.89)
- FlowTrack: Revisiting Optical Flow for Long-Range Dense Tracking (CVPR 2024 · similarity 0.89)
- StreamFlow: Streamlined Multi-Frame Optical Flow Estimation for Video Sequences (NeurIPS 2024 · similarity 0.89)
- Dense Optical Tracking: Connecting the Dots (CVPR 2024 · similarity 0.88)
- Taming generative video models for zero-shot optical flow extraction (NeurIPS 2025 · similarity 0.87)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
