Research map / Probabilistic and scientific modeling / Molecules and proteins
Iterative Residual Refinement for Joint Optical Flow and Occlusion 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 shared-weight IRR (ablation) | PWC-Net · Sintel Clean training | EPE: 3.13 -> 2.79 | Helped |
| Add occlusion estimation (ablation) | PWC-Net · FlyingChairs Full | EPE: 2.03 -> 1.94 | Helped |
| Add bidirectional flow estimation alone (ablation) | PWC-Net · FlyingChairs Full | EPE: 2.03 -> 2.06 | Hurt |
| Add IRR, bidirectional flow, and occlusion estimation together (ablation) | PWC-Net · Sintel Clean training | EPE: 3.13 -> 2.69 | Helped |
| Add bilateral refinement (ablation) | PWC-Net with IRR, bidirectional flow, and occlusion estimation · Sintel Clean training | EPE: 2.69 -> 2.32 | Helped |
| Add occlusion upsampling (ablation) | PWC-Net with bilateral refinement · Sintel Clean training | occlusion F1-score: 0.648 -> 0.677 | Helped |
| Add occlusion upsampling (ablation) | PWC-Net with bilateral refinement · Sintel Clean training | EPE: 2.32 -> 2.34 | Hurt |
| Replace the proposed upsampling layer with FlowNet2's refinement network (ablation) | PWC-Net with IRR, bidirectional flow, occlusion estimation, and bilateral refinement · Sintel Clean training | EPE: 2.32 -> 2.90 | Hurt |
| Use more shared-weight IRR iterations instead of stacking separate FlowNetS networks (ablation) | FlowNetS · Sintel Clean | EPE: 3.303 vs 3.391 | Helped |
| Use the full IRR-PWC design | PWC-Net · Sintel Final test | EPE: 5.04 -> 4.58 | Helped |
Closest papers by meaning
- OCAI: Improving Optical Flow Estimation by Occlusion and Consistency Aware Interpolation (CVPR 2024 · similarity 0.88)
- SKFlow: Learning Optical Flow with Super Kernels (NeurIPS 2022 · similarity 0.88)
- Occlusion Aware Unsupervised Learning of Optical Flow (CVPR 2018 · similarity 0.88)
- Recurrent Partial Kernel Network for Efficient Optical Flow Estimation (AAAI 2024 · similarity 0.87)
- DPFlow: Adaptive Optical Flow Estimation with a Dual-Pyramid Framework (CVPR 2025 · similarity 0.86)
- Context-Aware Iteration Policy Network for Efficient Optical Flow Estimation (AAAI 2024 · similarity 0.86)
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.