Research map / 3D vision / Depth and stereo
Robust Optical Flow in Rainy Scenes
Ruoteng Li, Robby T. Tan, Loong-Fah Cheong
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 the structure layer to the colored-residue image in the flow objective (ablation) | Ours-residue · Driving dataset rendered with rain streaks and accumulation | end point error | Improved |
| add the colored-residue image to the structure layer alone (ablation) | Ours-decomp · Driving dataset rendered with rain streaks and accumulation | end point error | Improved |
| use the colored-residue image alone under increasing rain accumulation (ablation) | Ours-residue · Driving dataset with rain accumulation | end point error | Hurt |
| preprocess with the deraining method of Yang et al. | Classic+NL-fast · Sintel (rain) | end point error: 7.97 -> 6.79 | Improved |
| preprocess with the deraining method of Yang et al. | FlowNet2 · Sintel (rain) | end point error: 7.68 -> 8.56 | Hurt |
| preprocess with the deraining method of Yang et al. | EpicFlow · Sintel (rain) | end point error: 14.92 -> 7.03 | Improved |
| preprocess with the deraining method of Yang et al. | DCFlow · FVR-660 | end point error: 46.71 -> 30.69 | Improved |
| preprocess with the deraining method of Yang et al. | Ours · Sintel | end point error: 5.46 -> 5.06 | No clear effect |
| fine-tune FlowNetS on synthetic-rain Flying Chairs | FlowNetS · Sintel (rain) | end point error: 7.40 -> 6.90 | Improved |
| fine-tune FlowNet2 on synthetic-rain Flying Chairs | FlowNet2 · Middlebury (rain) | end point error: 0.79 -> 1.24 | Hurt |
| fine-tune FlowNet2 on synthetic-rain Flying Chairs | FlowNet2 · FVR-660 (rain) | end point error: 5.73 -> 2.21 | Improved |
| run the proposed method on a static rainy scene instead of baselines | Ours · static real-rain image pair | average flow magnitude: 0.000195 pixel | Improved |
Closest papers by meaning
- RainFlow: Optical Flow Under Rain Streaks and Rain Veiling Effect (ICCV 2019 · similarity 0.90)
- Joint Bi-Layer Optimization for Single-Image Rain Streak Removal (ICCV 2017 · similarity 0.89)
- Self-Aligned Video Deraining With Transmission-Depth Consistency (CVPR 2021 · similarity 0.89)
- Adverse Weather Optical Flow: Cumulative Homogeneous-Heterogeneous Adaptation (TPAMI 2024 · similarity 0.88)
- Towards Universal Rainy Image Restoration: Benchmark and Baseline (AAAI 2025 · similarity 0.87)
- Boosting Image De-Raining via Central-Surrounding Synergistic Convolution (AAAI 2025 · 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.
