Research map / Image generation / Image restoration
MOERL: When Mixture-of-Experts Meet Reinforcement Learning for Adverse Weather Image Restoration
Tao Wang, Peiwen Xia, Bo Li, Peng-Tao Jiang, Zhe Kong, Kaihao Zhang, Tong Lu, Wenhan Luo
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 MOE layer (ablation) | residual-block baseline · Outdoor-Rain | PSNR (dB): 28.89 -> 31.48 | Improved |
| add Transformer layer (ablation) | residual-block baseline · Outdoor-Rain | PSNR (dB): 28.89 -> 30.21 | Improved |
| add Transformer layer to MOE layer (ablation) | MOERL network · Outdoor-Rain | PSNR (dB): 31.48 -> 32.70 | Improved |
| add channel experts to spatial experts (ablation) | MOE layer · Outdoor-Rain | PSNR (dB): 32.20 -> 32.70 | Improved |
| add spatial experts to channel experts (ablation) | MOE layer · Outdoor-Rain | PSNR (dB): 32.03 -> 32.70 | Improved |
| add routing prompt module (ablation) | MOERL · Outdoor-Rain | PSNR (dB): 31.12 -> 32.79 | Improved |
| add RL training (ablation) | MOERL · Outdoor-Rain | PSNR (dB): 32.40 -> 32.70 | Improved |
| add critic network (ablation) | RL strategy · Outdoor-Rain | PSNR (dB): 32.54 -> 32.70 | Improved |
| add quality prompt to reward model (ablation) | Q-Align reward · Outdoor-Rain | PSNR (dB): 32.68 -> 32.70 | Improved |
| add degradation prompt to reward model (ablation) | Q-Align reward · Outdoor-Rain | PSNR (dB): 32.62 -> 32.70 | Improved |
| replace LLaVA with Q-Align reward (ablation) | MOERL RL · average of 4 sets | PSNR (dB): 33.72 -> 34.23 | Improved |
| replace InternVL with Q-Align reward (ablation) | MOERL RL · average of 4 sets | PSNR (dB): 33.90 -> 34.23 | Improved |
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.
- Restormer — Restormer: Efficient Transformer for High-Resolution Image Restoration (CVPR 2022) · better on 4 of 4 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
- Real-World Adverse Weather Image Restoration via Dual-Level Reinforcement Learning with High-Quality Cold Start (NeurIPS 2025 · similarity 0.91)
- Complexity Experts are Task-Discriminative Learners for Any Image Restoration (CVPR 2025 · similarity 0.87)
- Continuous Adverse Weather Removal via Degradation-Aware Distillation (CVPR 2025 · similarity 0.87)
- AWRaCLe: All-Weather Image Restoration Using Visual In-Context Learning (AAAI 2025 · similarity 0.87)
- Language-driven All-in-one Adverse Weather Removal (CVPR 2024 · similarity 0.86)
- Multi-axis Prompt and Multi-dimension Fusion Network for All-in-one Weather-degraded Image Restoration (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.
