Research map / Image generation / Image restoration
Learning to Adapt to Light
Kai-Fu Yang, Cheng Cheng, Shixuan Zhao, Hong-Mei Yan, Xian-Shi Zhang, Yong-Jie Li
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 frequency decomposition with a high-frequency pathway (two-pathway vs one-pathway) (ablation) | LA-Net · LOL test | PSNR: 15.570 -> 21.71 | Improved |
| add frequency decomposition with a high-frequency pathway (two-pathway vs one-pathway) (ablation) | LA-Net · LOL test | SSIM: 0.377 -> 0.805 | Improved |
| add frequency decomposition with a high-frequency pathway (two-pathway vs one-pathway) (ablation) | LA-Net · Exposure correction | PSNR: 17.495 -> 20.704 | Improved |
| add frequency decomposition with a high-frequency pathway (two-pathway vs one-pathway) (ablation) | LA-Net · Exposure correction | SSIM: 0.745 -> 0.819 | Improved |
| add frequency decomposition with a high-frequency pathway (two-pathway vs one-pathway) (ablation) | LA-Net · HDRPS tone mapping | TMQI: 0.8975 -> 0.8803 | Hurt |
| increase the number of NR curves from 4 to 16 (ablation) | LA-Net · LOL test | PSNR: 21.575 -> 21.713 | No clear effect |
| increase the number of NR curves from 16 to 32 (ablation) | LA-Net · LOL test | PSNR: 21.713 -> 21.580 | No clear effect |
| increase the number of NR curves from 4 to 32 (runtime cost) (ablation) | LA-Net · LOL test | Time (ms): 30 -> 110 | Hurt |
| use LA-Net instead of Afifi et al. for exposure correction | LA-Net · Exposure correction | PSNR: 19.377 -> 20.704 | Improved |
| use LA-Net instead of KinD++ for low-light enhancement | LA-Net · LOL test | NIQE: 3.88 -> 3.10 | Improved |
| use LA-Net instead of KinD++ for low-light enhancement | LA-Net · LOL test | SSIM: 0.82 -> 0.81 | Hurt |
Closest papers by meaning
- Learning Adaptive Lighting via Channel-Aware Guidance (ICML 2025 · similarity 0.88)
- Empowering Low-Light Image Enhancer through Customized Learnable Priors (ICCV 2023 · similarity 0.86)
- Attention Guided Low-Light Image Enhancement with a Large Scale Low-Light Simulation Dataset (IJCV 2021 · similarity 0.86)
- Unsupervised Illumination Adaptation for Low-Light Vision (TPAMI 2024 · similarity 0.86)
- Improving Visual and Downstream Performance of Low-Light Enhancer with Vision Foundation Models Collaboration (CVPR 2025 · similarity 0.85)
- Interpretable Unsupervised Joint Denoising and Enhancement for Real-World low-light Scenarios (ICLR 2025 · 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.
