Research map / Architectures and efficiency / CNN design
Learning Optical Flow With Kernel Patch Attention
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 local kernel patch attention to RAFT | RAFT · KITTI-15 val, C + T | EPE: 4.69 → 4.46 | Helped |
| Add local kernel patch attention to RAFT | RAFT · Sintel val, C + T, clean | EPE: 1.43 → 1.28 | Helped |
| Add local kernel patch attention to RAFT | RAFT · Sintel val, C + T, final | EPE: 2.71 → 2.68 | Helped |
| Add local kernel patch attention to RAFT | RAFT · Sintel test, clean | EPE: 1.39 → 1.35 | Helped |
| Add local kernel patch attention to RAFT | RAFT · Sintel test, final | EPE: 2.47 → 2.36 | Helped |
| Add local kernel patch attention to RAFT | RAFT · KITTI-15 test | F1-all: 5.10 → 4.60 | Helped |
| Use patch size 5 rather than 9 (ablation) | KPA-Flow · KITTI-15 val, C + T | EPE: 4.46 → 4.72 | Hurt |
| Use patch size 7 rather than 9 (ablation) | KPA-Flow · KITTI-15 val, C + T | EPE: 4.46 → 4.58 | Hurt |
| Use patch size 11 rather than 9 (ablation) | KPA-Flow · KITTI-15 val, C + T | EPE: 4.46 → 4.49 | Hurt |
| Use patch size 13 rather than 9 (ablation) | KPA-Flow · KITTI-15 val, C + T | EPE: 4.46 → 4.51 | Hurt |
| Use kernel shape 1 × 1 rather than 3 × 3 (ablation) | KPA-Flow · KITTI-15 val, C + T | EPE: 4.46 → 4.91 | Hurt |
| Use kernel shape 5 × 5 rather than 3 × 3 (ablation) | KPA-Flow · KITTI-15 val, C + T | EPE: 4.46 → 4.57 | Hurt |
Closest papers by meaning
- SKFlow: Learning Optical Flow with Super Kernels (NeurIPS 2022 · similarity 0.88)
- Recurrent Partial Kernel Network for Efficient Optical Flow Estimation (AAAI 2024 · similarity 0.88)
- CNN-Based Patch Matching for Optical Flow With Thresholded Hinge Embedding Loss (CVPR 2017 · similarity 0.87)
- Taming generative video models for zero-shot optical flow extraction (NeurIPS 2025 · similarity 0.85)
- FlowFeat: Pixel-Dense Embedding of Motion Profiles (NeurIPS 2025 · similarity 0.85)
- Injecting Frame-Event Complementary Fusion into Diffusion for Optical Flow in Challenging Scenes (NeurIPS 2025 · similarity 0.85)
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.