Research map / Image generation / Fast diffusion sampling
FlowPainter: Inpainting Optical Flow via Confidence-Guided Completion
Yuang Meng, Chenyang Wu, Xianshun Liu, Chun-Le Guo, Zichen Liang, Lina Lei, Jie Liang, Hui Zeng, Chongyi Li, Lei Zhang
Read the paperCheck 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 |
|---|---|---|---|
| add occlusion mask supervision to the confidence mask (with hard-flow mask) (ablation) | FlowPainter · Sintel (train) final | EPE: 1.37 -> 1.32 | Improved |
| add occlusion mask supervision to the confidence mask (with hard-flow mask) (ablation) | FlowPainter · KITTI-15 (train) | Fl-all: 7.01 -> 6.84 | Improved |
| add hard-flow mask supervision to the confidence mask (with occlusion mask) (ablation) | FlowPainter · Sintel (train) final | EPE: 1.95 -> 1.32 | Improved |
| add hard-flow mask supervision to the confidence mask (with occlusion mask) (ablation) | FlowPainter · KITTI-15 (train) | Fl-all: 8.13 -> 6.84 | Improved |
| add both occlusion and hard-flow masks vs neither (ablation) | FlowPainter · Sintel (train) final | EPE: 2.17 -> 1.32 | Improved |
| add time step decay module (with inpainting guidance) (ablation) | FlowPainter · Sintel (train) final | EPE: 2.32 -> 1.32 | Improved |
| add time step decay module (with inpainting guidance) (ablation) | FlowPainter · KITTI-15 (train) | Fl-all: 10.12 -> 6.84 | Improved |
| add confidence-mask inpainting guidance (with time step decay) (ablation) | FlowPainter · Sintel (train) final | EPE: 1.67 -> 1.32 | Improved |
| add confidence-mask inpainting guidance (with time step decay) (ablation) | FlowPainter · KITTI-15 (train) | Fl-all: 8.97 -> 6.84 | Improved |
| add both time step decay and inpainting guidance vs neither (ablation) | FlowPainter · Sintel (train) final | EPE: 2.74 -> 1.32 | Improved |
| lower initialization threshold from 0.5 to 0.3 (ablation) | FlowPainter · Sintel (train) final | EPE: 1.32 -> 1.78 | Hurt |
| raise initialization threshold from 0.5 to 0.7 (ablation) | FlowPainter · KITTI-15 (train) | Fl-all: 6.84 -> 8.71 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- FlowDiffuser — FlowDiffuser: Advancing Optical Flow Estimation with Diffusion Models (CVPR 2024) · better on 9 of 9 shared results
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
- FlowDiffuser: Advancing Optical Flow Estimation with Diffusion Models (CVPR 2024 · similarity 0.91)
- DifFlow3D: Toward Robust Uncertainty-Aware Scene Flow Estimation with Iterative Diffusion-Based Refinement (CVPR 2024 · similarity 0.90)
- DiffSF: Diffusion Models for Scene Flow Estimation (NeurIPS 2024 · similarity 0.89)
- Hierarchical Flow Diffusion for Efficient Frame Interpolation (CVPR 2025 · similarity 0.89)
- Neuroexplicit Diffusion Models for Inpainting of Optical Flow Fields (ICML 2024 · similarity 0.89)
- Video Diffusion Models Are Strong Video Inpainter (AAAI 2025 · similarity 0.88)
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.
