Research map / Image generation / Image restoration
Towards a Universal Image Degradation Model via Content-Degradation Disentanglement
Wenbo Yang, Zhongling Wang, Zhou Wang
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 entropy regularization (disentangle-by-compression) (ablation) | degradation model · FFHQ-X Mixed transfer | LPIPS: .334 -> .286 | Improved |
| add entropy regularization (disentangle-by-compression) (ablation) | degradation model · FFHQ-X Direct transfer | LPIPS: .290 -> .271 | Improved |
| add IDA layer (ablation) | degradation model · FFHQ-X Mixed transfer | LPIPS: .356 -> .286 | Improved |
| add IDA layer (ablation) | degradation model · FFHQ-X Direct transfer | LPIPS: .298 -> .271 | Improved |
| add IDA layer (ablation) | degradation model · FFHQ-X Global-only transfer | LPIPS: .404 -> .388 | Improved |
| add IDEN on top of IDA-less model (ablation) | degradation model · FFHQ-X Direct transfer | LPIPS: .577 -> .298 | Improved |
| add IDEN on top of IDA-less model (ablation) | degradation model · FFHQ-X Global-only transfer | LPIPS: .394 -> .404 | No clear effect |
| plug our degradation model into blind RSG GAN inversion | RSG · FFHQ-X NA | accuracy (LPIPS): .680 -> .513 | Improved |
| plug our degradation model into blind RSG GAN inversion | RSG · FFHQ-X UNAP | realism (pFID): 147.4 -> 31.7 | Improved |
| plug our degradation model into blind RSG GAN inversion | RSG · FFHQ-X UA | accuracy (LPIPS): .519 -> .512 | No clear effect |
| transfer degradation instead of reproduce | degradation model · synthetic WQI | LPIPS: 0.295 -> 0.306 | No clear effect |
| use universal model instead of specialized film grain baseline | degradation model · FilmGrainStyle | reproduction score | Improved |
Closest papers by meaning
- UniRes: Universal Image Restoration for Complex Degradations (ICCV 2025 · similarity 0.90)
- Unsupervised Diffusion-Based Degradation Modeling for Real-World Super-Resolution (AAAI 2025 · similarity 0.89)
- Visual-Instructed Degradation Diffusion for All-in-One Image Restoration (CVPR 2025 · similarity 0.88)
- GenDeg: Diffusion-based Degradation Synthesis for Generalizable All-In-One Image Restoration (CVPR 2025 · similarity 0.88)
- Unsupervised Degradation Representation Aware Transform for Real-World Blind Image Super-Resolution (AAAI 2025 · similarity 0.88)
- Degradation-Aware Feature Perturbation for All-in-One Image Restoration (CVPR 2025 · 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.
