Research map / Architectures and efficiency / Neural architecture search
Learning With Nested Scene Modeling and Cooperative Architecture Search for Low-Light Vision
Risheng Liu, Long Ma, Tengyu Ma, Xin Fan, Zhongxuan 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 noise estimation (RUAS_A over RUAS) (ablation) | RUAS · LOL | PSNR: 18.226 -> 19.751 | Improved |
| add noise estimation (RUAS_A over RUAS) (ablation) | RUAS · LOL | LPIPS: 0.354 -> 0.393 | Hurt |
| add noise removal (RUAS over RUAS_S) (ablation) | RUAS_S · LOL | PSNR: 16.038 -> 18.226 | Improved |
| use cooperative search instead of global search (ablation) | RUAS · MIT-Adobe 5K | PSNR: 14.496 -> 18.226 | Improved |
| use cooperative search instead of independent search (ablation) | RUAS · MIT-Adobe 5K | PSNR: 15.841 -> 18.226 | Improved |
| use searched architecture instead of supernet (ablation) | RUAS · MIT-Adobe 5K | PSNR: 18.940 -> 21.018 | Improved |
| use searched architecture instead of single 3-2-RDC convolution (ablation) | RUAS · MIT-Adobe 5K | PSNR: 19.391 -> 21.018 | Improved |
| increase scene stages K from 1 to 3 (ablation) | RUAS · MIT-Adobe 5K | PSNR: 13.013 -> 21.018 | Improved |
| increase scene stages K from 3 to 5 (ablation) | RUAS · MIT-Adobe 5K | PSNR: 21.018 -> 20.846 | Hurt |
| use hierarchical (pretrain+finetune) instead of end-to-end training (ablation) | RUAS · DARK FACE | mAP: 0.685 -> 0.693 | Improved |
| add scene loss to end-to-end training (ablation) | RUAS · DARK FACE | mAP: 0.685 -> 0.660 | Hurt |
| add scene module to task module (ablation) | SSD task module · DARK FACE | mAP: 0.611 -> 0.639 | Improved |
Closest papers by meaning
- Retinex-Inspired Unrolling With Cooperative Prior Architecture Search for Low-Light Image Enhancement (CVPR 2021 · similarity 0.92)
- From Enhancement to Understanding: Build a Generalized Bridge for Low-light Vision via Semantically Consistent Unsupervised Fine-tuning (ICCV 2025 · similarity 0.87)
- Semantically Contrastive Learning for Low-Light Image Enhancement (AAAI 2022 · similarity 0.87)
- Unsupervised Illumination Adaptation for Low-Light Vision (TPAMI 2024 · similarity 0.87)
- Interpretable Unsupervised Joint Denoising and Enhancement for Real-World low-light Scenarios (ICLR 2025 · similarity 0.85)
- UniLumos: Fast and Unified Image and Video Relighting with Physics-Plausible Feedback (NeurIPS 2025 · 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.
