Research map / Image generation / Image restoration
Depth-Aware Blind Image Decomposition for Real-World Adverse Weather Recovery
Chao Wang, Zhedong Zheng, Ruijie Quan, Yi Yang
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 |
|---|---|---|---|
| replace the DBA depth-aware attention module with a 1x1 convolution (ablation) | DeBNet · BID-CityScapes case 5 | PSNR: 27.50 -> 28.18 | Improved |
| add ECA channel attention module (ablation) | DeBNet · BID-CityScapes case 5 | PSNR: 27.95 -> 28.18 | Improved |
| add DBA module (ablation) | DeBNet · BID-CityScapes case 1 | PSNR: 28.87 -> 29.51 | Improved |
| add neural architecture search to the encoder (ablation) | DeBNet · BID-CityScapes case 5 | PSNR: 27.88 -> 28.18 | Improved |
| add neural architecture search to the encoder (ablation) | DeBNet · BID-CityScapes case 2 (simple) | PSNR: 29.79 -> 29.72 | No clear effect |
| vary GAN loss weight lambda_s between 0.05 and 0.5 (ablation) | DeBNet · BID-CityScapes case 2 | PSNR: 29.70 -> 29.68 | No clear effect |
| train on depth-aware BID-CityScapes instead of the BIDeN dataset (ablation) | DeBNet · Snow100k-M | PSNR: 31.15 -> 33.80 | Improved |
| increase NAS L1 penalty to lambda=0.05 (ablation) | DeBNet · BID-CityScapes case 5 | PSNR: 28.18 -> 24.91 | Hurt |
| pre-train on BID-CityScapes then fine-tune on target dataset (ablation) | DeBNet · Deraindrop / Snow100k-M / SICE | PSNR: 33.05 -> 33.11 on Deraindrop | Improved |
| use DeBNet with depth guidance and NAS instead of diffusion-based WeatherDiff | BID-CityScapes case 1 · BID-CityScapes case 1 | PSNR: 28.11 -> 29.51 | Improved |
Closest papers by meaning
- Weather-Conditioned Depth Anything (ECCV 2026 · similarity 0.89)
- Depth-Attentional Features for Single-Image Rain Removal (CVPR 2019 · similarity 0.89)
- Blind Image Decomposition (ECCV 2022 · similarity 0.88)
- Synthetic-to-Real Self-supervised Robust Depth Estimation via Learning with Motion and Structure Priors (CVPR 2025 · similarity 0.86)
- Real-World Adverse Weather Image Restoration via Dual-Level Reinforcement Learning with High-Quality Cold Start (NeurIPS 2025 · similarity 0.85)
- Genuine Knowledge from Practice: Diffusion Test-Time Adaptation for Video Adverse Weather Removal (CVPR 2024 · similarity 0.85)
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.
