Research map / Image generation / Image restoration
Teaching Tailored to Talent: Adverse Weather Restoration via Prompt Pool and Depth-Anything Constraint
Sixiang Chen, Tian Ye, Kai Zhang, Zhaohu Xing, Yunlong Lin, Lei Zhu
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 prompt pool for weather prompts (ablation) | T3-DiffWeather · Outdoor-Rain | PSNR: 31.05 -> 31.99 | Improved |
| use prompt pool instead of shared learnable prompts (ablation) | T3-DiffWeather · Outdoor-Rain | PSNR: 31.38 -> 31.99 | Improved |
| add matched keys for sub-prompt selection (ablation) | prompt pool · Outdoor-Rain | PSNR: 31.72 -> 31.99 | Improved |
| add general prompts for scene modeling (ablation) | T3-DiffWeather · Outdoor-Rain | PSNR: 31.52 -> 31.99 | Improved |
| constrain general prompts with Depth-Anything features (ablation) | general prompts · Outdoor-Rain | PSNR: 31.67 -> 31.99 | Improved |
| use Depth-Anything instead of DINOv2 (ablation) | general prompts · Outdoor-Rain | PSNR: 31.82 -> 31.99 | Improved |
| add contrastive prompt loss (ablation) | T3-DiffWeather · Outdoor-Rain | PSNR: 31.71 -> 31.99 | Improved |
| scale Depth-Anything from ViT-S to ViT-L (ablation) | general prompt constraint · Outdoor-Rain | PSNR: 31.99 -> 32.06 | No clear effect |
| increase general prompt length from 256 to 320 (ablation) | general prompts · Outdoor-Rain | PSNR: 31.99 -> 32.03 | No clear effect |
| increase sub-prompt length from 64 to 128 (ablation) | prompt pool · Outdoor-Rain | PSNR: 31.99 -> 32.04 | No clear effect |
| reconstruct degradation residual with 2 sampling steps instead of WeatherDiffusion 25 steps | diffusion restoration · 256x256 complexity | GFLOPs: 248.4G x 25 steps -> 59.82G x 2 steps | Improved |
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)
- Genuine Knowledge from Practice: Diffusion Test-Time Adaptation for Video Adverse Weather Removal (CVPR 2024 · similarity 0.88)
- Continuous Adverse Weather Removal via Degradation-Aware Distillation (CVPR 2025 · similarity 0.88)
- Multi-axis Prompt and Multi-dimension Fusion Network for All-in-one Weather-degraded Image Restoration (AAAI 2025 · similarity 0.88)
- AWRaCLe: All-Weather Image Restoration Using Visual In-Context Learning (AAAI 2025 · similarity 0.87)
- Harnessing Joint Rain-/Detail-aware Representations to Eliminate Intricate Rains (ICLR 2024 · similarity 0.87)
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.
