Research map / Label-efficient and robust learning / Self-supervised and contrastive
Imposing Consistency for Optical Flow Estimation
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 occlusion and transformation consistency to RAFT-small | RAFT-small · Sintel train Clean | EPE: 2.21 -> 1.95 | Helped |
| Add occlusion and transformation consistency to RAFT-small | RAFT-small · KITTI train | Fl-all: 26.9 -> 22.1 | Helped |
| Add occlusion and transformation consistency to RAFT | RAFT · KITTI test | Fl-all: 5.10 -> 4.72 | Helped |
| Fine-tune with KITTI-2012 and KITTI-2015 together | RAFT + OCTC · KITTI test | Fl-all: 4.72 -> 4.33 | Helped |
| Use zero forcing on identical image pairs (It, It) (ablation) | RAFT-small · Sintel train Clean | EPE: 2.21 -> 2.23 | Hurt |
| Use zero forcing on identical image pairs (It, It) (ablation) | RAFT-small · KITTI-15 train | Fl-all: 26.9 -> 25.8 | Helped |
| Add zero forcing using a generated occlusion pair (ablation) | RAFT-small · Sintel train Clean | EPE: 2.21 -> 2.17 | Helped |
| Add mask-match loss (ablation) | RAFT-small · Sintel train Clean | EPE: 2.21 -> 2.11 | Helped |
| Combine zero forcing on generated occlusions with mask-match loss (ablation) | RAFT-small · Sintel train Clean | EPE: 2.11 -> 2.05 | Helped |
| Add transformation consistency with horizontal flips (ablation) | RAFT-small · KITTI-15 train | Fl-all: 26.9 -> 22.6 | Helped |
| Add transformation consistency with random rotations (ablation) | RAFT-small · Sintel train Final | EPE: 3.35 -> 3.15 | Helped |
| Extend frame hopping from k=1,2 to k=1,2,3 (ablation) | RAFT-small + rotation TC · KITTI-15 train | Fl-epe: 6.50 -> 6.69 | Hurt |
Closest papers by meaning
- OCAI: Improving Optical Flow Estimation by Occlusion and Consistency Aware Interpolation (CVPR 2024 · similarity 0.91)
- Learning by Analogy: Reliable Supervision From Transformations for Unsupervised Optical Flow Estimation (CVPR 2020 · similarity 0.90)
- DPFlow: Adaptive Optical Flow Estimation with a Dual-Pyramid Framework (CVPR 2025 · similarity 0.87)
- ADFactory: An Effective Framework for Generalizing Optical Flow with NeRF (CVPR 2024 · similarity 0.87)
- UnSAMFlow: Unsupervised Optical Flow Guided by Segment Anything Model (CVPR 2024 · similarity 0.87)
- Recurrent Partial Kernel Network for Efficient Optical Flow Estimation (AAAI 2024 · 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.