Research map / Image generation / Image restoration
Difficulty-Conditioned Attribute-Specific Restoration for Low-Light Image Enhancement
Mingzhu Zhang, Shuang Li, Jiaxu Leng, Long Sun, Can Zhang, Miaoqing Wang, Xinbo Gao
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 GIM to baseline (ablation) | U-Net/CIDNet-style baseline · LOL-v2-real | PSNR: 23.498 -> 23.625 | Improved |
| add SRM to baseline (ablation) | baseline · LOL-v2-real | PSNR: 23.498 -> 23.765 | Improved |
| add both GIM and SRM (ablation) | baseline · LOL-v2-real | PSNR: 23.498 -> 23.894 | Improved |
| add adaptive prior-weighted loss Lapw (ablation) | DCASR with GIM and SRM · LOL-v2-real | PSNR: 23.894 -> 24.053 | Improved |
| use GIM in I branch and SRM in HV branch instead of channel attention (ablation) | DCASR · LOL-v2-real | PSNR: 23.012 -> 23.894 | Improved |
| place GIM in I branch and SRM in HV branch rather than swapped (ablation) | DCASR · LOL-v2-real | PSNR: 22.662 -> 23.894 | Improved |
| use heterogeneous GIM+SRM rather than dual GIM (ablation) | DCASR · LOL-v2-real | PSNR: 22.923 -> 23.894 | Improved |
| use heterogeneous GIM+SRM rather than dual SRM (ablation) | DCASR · LOL-v2-real | PSNR: 22.241 -> 23.894 | Improved |
| use diffusion-predicted difficulty prior instead of no prior (ablation) | DCASR · LOL-v2-real | PSNR: 23.498 -> 24.053 | Improved |
| use residual-derived diffusion prior instead of confidence map guidance (ablation) | DCASR · LOL-v2-real | PSNR: 23.768 -> 24.053 | Improved |
| use SNR map as prior guidance instead of no prior (ablation) | DCASR · LOL-v2-real | PSNR: 23.498 -> 23.129 | Hurt |
| use diffusion to predict prior instead of transformer (ablation) | DCASR · LOL-v2-real | PSNR: 23.611 -> 24.053 | Improved |
Closest papers by meaning
- Learnable Feature Patches and Vectors for Boosting Low-light Image Enhancement without External Knowledge (ICCV 2025 · similarity 0.90)
- Zero-Shot Low-Light Image Enhancement via Latent Diffusion Models (AAAI 2025 · similarity 0.90)
- Interpretable Unsupervised Joint Denoising and Enhancement for Real-World low-light Scenarios (ICLR 2025 · similarity 0.87)
- AGLLDiff: Guiding Diffusion Models Towards Unsupervised Training-free Real-world Low-light Image Enhancement (AAAI 2025 · similarity 0.87)
- Luminance-Aware Statistical Quantization: Unsupervised Hierarchical Learning for Illumination Enhancement (NeurIPS 2025 · similarity 0.87)
- Zero-Reference Low-Light Enhancement via Physical Quadruple Priors (CVPR 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.
