Research map / Video / Video segmentation and tracking
Learning Large Motion Estimation from Intermediate Representations with a High-Resolution Optical Flow Dataset Featuring Long-Range Dynamic Motion
Hoonhee Cho, Yuhwan Jeong, Kuk-Jin Yoon
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 |
|---|---|---|---|
| Add the Relay training strategy to RAFT | RAFT · RelayFlow-4K test | EPE (all regions): 18.54 -> 9.81 | Improved |
| Add the Relay training strategy to GMA | GMA · RelayFlow-4K test | EPE (all regions): 22.08 -> 7.96 | Improved |
| Add the Relay strategy to GMFlowNet | GMFlowNet · RelayFlow-4K test | EPE (all regions): 25.24 -> 13.24 | Improved |
| Add the Relay strategy to GMFlow | GMFlow · RelayFlow-4K test | EPE (all regions): 15.44 -> 8.92 | Improved |
| Add Relay to RAFT, long-range flows | RAFT · RelayFlow-4K test | EPE s160+: 466.83 -> 230.32 | Improved |
| Add Relay to RAFT, unmatched regions | RAFT · RelayFlow-4K test | EPE unmatched: 223.82 -> 154.83 | Improved |
| Add Relay to GMFlow, short-range flows | GMFlow · RelayFlow-4K test | EPE s0-40: 1.99 -> 2.28 | Hurt |
| Add Ldis alone to RAFT (ablation) | RAFT · RelayFlow-4K test | EPE (all regions): 18.54 -> 15.36 | Improved |
| Add all of Ldis, incremental time-step learning and LKD (ablation) | RAFT · RelayFlow-4K test | EPE (all regions): 18.54 -> 9.81 | Improved |
| Add Ldis to incremental learning plus LKD (ablation) | RAFT · RelayFlow-4K test | EPE (all regions): 11.16 -> 9.81 | Improved |
| Mask the cost volume with the match map during aggregation (ablation) | RAFT-Relay · RelayFlow-4K test | EPE (all regions): 12.10 -> 9.81 | Improved |
| Train through successive incremental stages 0 to 4 (ablation) | RAFT-Relay · RelayFlow-4K test | EPE (all regions): 14.04 -> 9.81 | Improved |
Closest papers by meaning
- DPFlow: Adaptive Optical Flow Estimation with a Dual-Pyramid Framework (CVPR 2025 · similarity 0.89)
- FlowTrack: Revisiting Optical Flow for Long-Range Dense Tracking (CVPR 2024 · similarity 0.88)
- StreamFlow: Streamlined Multi-Frame Optical Flow Estimation for Video Sequences (NeurIPS 2024 · similarity 0.88)
- MemFlow: Optical Flow Estimation and Prediction with Memory (CVPR 2024 · similarity 0.88)
- Sparse Global Matching for Video Frame Interpolation with Large Motion (CVPR 2024 · similarity 0.87)
- EDCFlow: Exploring Temporally Dense Difference Maps for Event-based Optical Flow Estimation (CVPR 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.
