Research map / Image generation / Image restoration
Robust Representation Learning With Feedback for Single Image Deraining
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 |
|---|---|---|---|
| Use RLNet instead of the strongest listed baseline | single-image deraining · DID-data | PSNR: 30.40 -> 32.62 | Helped |
| Use RLNet instead of the strongest listed baseline | single-image deraining · Rain200H | PSNR: 28.83 -> 28.87 | Helped |
| Use RLNet instead of the strongest listed baseline | single-image deraining · Rain100L | PSNR: 38.60 -> 37.38 | Hurt |
| Use RLNet instead of the strongest listed baseline | single-image deraining · Rain800 | PSNR: 25.52 -> 27.95 | Helped |
| Add the error detector to the base network (ablation) | Rain200H deraining, M4 · Rain200H | PSNR: 27.39 -> 28.69 | Helped |
| Add error-detection loss Le2 (ablation) | Rain200H deraining, M5 · Rain200H | PSNR: 27.40 -> 28.69 | Helped |
| Add feature-compensation losses Lc and Lp (ablation) | Rain200H deraining, M6 · Rain200H | PSNR: 28.69 -> 28.87 | Helped |
| Add FFRB in place of the ordinary residual block (ablation) | Rain200H deraining, M1 · Rain200H | PSNR: 26.91 -> 27.13 | Helped |
| Replace single-stream with multi-stream architecture (ablation) | Rain200H deraining, M2 · Rain200H | PSNR: 27.13 -> 27.42 | Helped |
| Use a non-absolute error map for direct compensation (ablation) | RLNet- · Rain200H | PSNR: 27.70 vs 27.57 | Hurt |
| Set error-detector threshold θ1 away from 0.05 (ablation) | RLNet · Rain200H | PSNR: 28.48; 28.57; 28.61 vs 28.87 | Hurt |
| Set feature-compensation transformation θ2 away from 0.15 (ablation) | RLNet · Rain200H | PSNR: 28.74; 28.81; 28.85 vs 28.87 | Hurt |
Closest papers by meaning
- Multi-Decoding Deraining Network and Quasi-Sparsity Based Training (CVPR 2021 · similarity 0.88)
- Unsupervised Deraining: Where Contrastive Learning Meets Self-Similarity (CVPR 2022 · similarity 0.88)
- Channel Consistency Prior and Self-Reconstruction Strategy Based Unsupervised Image Deraining (CVPR 2025 · similarity 0.86)
- Bidirectional Multi-Scale Implicit Neural Representations for Image Deraining (CVPR 2024 · similarity 0.85)
- Rethinking Multi-Scale Representations in Deep Deraining Transformer (AAAI 2024 · similarity 0.85)
- Harnessing Joint Rain-/Detail-aware Representations to Eliminate Intricate Rains (ICLR 2024 · similarity 0.84)
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.