Research map / Image generation / Image restoration
CFSNet: Toward a Controllable Feature Space for Image Restoration
Wei Wang, Ruiming Guo, Yapeng Tian, Wenming 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 |
|---|---|---|---|
| add the learned coupling-coefficient mapping network (vs same alpha for all channels and layers, CFSNet-SA) (ablation) | CFSNet · BSD68 denoising, unseen noise level 30 | best PSNR | Improved |
| add the learned coupling-coefficient mapping network (vs CFSNet-SA) (ablation) | CFSNet · LIVE1 JPEG deblocking, unseen quality 20 | best PSNR | Improved |
| use CFSNet interpolation instead of FFDNet at unseen noise level | CFSNet vs FFDNet · BSD68 sigma=30 | PSNR: 28.39 -> 28.39 | No clear effect |
| use CFSNet instead of DnCNN-3 on unseen JPEG quality 20 | CFSNet · LIVE1 JPEG deblocking | PSNR: 31.59 -> 31.71 | Improved |
| use CFSNet instead of DnCNN-3 on unseen JPEG quality 30 | CFSNet · LIVE1 JPEG deblocking | PSNR: 32.98 -> 33.16 | Improved |
| use CFSNet instead of IRCNN for super-resolution with unseen degradation | CFSNet · Urban100 | PSNR: 26.77 -> 27.33 | Improved |
| use CFSNet instead of FFDNet on unseen noise level 15 | CFSNet vs FFDNet · BSD68 sigma=15 | PSNR: 31.63 -> 31.29 | Hurt |
| use CFSNet instead of FFDNet at endpoint sigma=50 | CFSNet vs FFDNet · BSD68 sigma=50 | PSNR: 26.29 -> 26.28 | No clear effect |
Later papers that compared with this method
1 later accepted paper reports CFSNet as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.
- Tunable Convolutions With Parametric Multi-Loss Optimization (CVPR 2023 · better on 5 of 6 shared results)
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Learning Enriched Features for Fast Image Restoration and Enhancement (TPAMI 2022 · similarity 0.87)
- Learning Enriched Features for Real Image Restoration and Enhancement (ECCV 2020 · similarity 0.87)
- xUnit: Learning a Spatial Activation Function for Efficient Image Restoration (CVPR 2018 · similarity 0.86)
- Path-Restore: Learning Network Path Selection for Image Restoration (TPAMI 2021 · similarity 0.86)
- Omni-Kernel Network for Image Restoration (AAAI 2024 · similarity 0.85)
- Efficient Visual State Space Model for Image Deblurring (CVPR 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.
