Research map / Image generation / Image restoration
UniLDiff: Unlocking the Power of Diffusion Priors for All-in-One Image Restoration
Zihan Cheng, Liangtai Zhou, Dian Chen, Ni Tang, Xiaotong Luo, Yuan Xie, Yanyun Qu
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 DAFF fusion (vs no fusion) (ablation) | UniLDiff · 5-task average | PSNR: 23.12 -> 27.14 | Improved |
| use full DAFF instead of residual self-attention fusion (ablation) | UniLDiff · 5-task average | PSNR: 24.08 -> 27.14 | Improved |
| combine double-stream and single-stream paths (ablation) | UniLDiff · 5-task average | MUSIQ: 51.02 -> 61.35 | Improved |
| add DAEM on top of DAFF and task prompt (ablation) | UniLDiff · 5-task average | PSNR: 27.36 -> 30.27 | Improved |
| add task prompt on top of DAFF (ablation) | UniLDiff · 5-task average | PSNR: 27.14 -> 27.36 | Improved |
| add task prompt on top of DAFF + DAEM (ablation) | UniLDiff · 5-task average | PSNR: 30.21 -> 30.27 | No clear effect |
| add task prompt alone (no DAFF) (ablation) | UniLDiff · 5-task average | MUSIQ: 41.31 -> 50.12 | Improved |
| diffusion prior UniLDiff vs non-diffusion DFPIR | UniLDiff · 3-task average | PSNR: 32.75 -> 32.18 | Hurt |
| diffusion prior UniLDiff vs DA-RCOT on perceptual no-reference metrics | UniLDiff · 3-task average | MUSIQ: 67.82 -> 68.89 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- Unirestore — UniRestore: Unified Perceptual and Task-Oriented Image Restoration Model Using Diffusion Prior (CVPR 2025) · better on 2 of 2 shared results
- InstructIR — InstructIR: High-Quality Image Restoration Following Human Instructions (ECCV 2024) · better on 1 of 1 shared result
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
- Diffusion Once and Done: Degradation-Aware LoRA for All-in-One Image Restoration (AAAI 2026 · similarity 0.91)
- UniRes: Universal Image Restoration for Complex Degradations (ICCV 2025 · similarity 0.90)
- UP-Restorer: When Unrolling Meets Prompts for Unified Image Restoration (AAAI 2025 · similarity 0.89)
- Visual-Instructed Degradation Diffusion for All-in-One Image Restoration (CVPR 2025 · similarity 0.89)
- GenDeg: Diffusion-based Degradation Synthesis for Generalizable All-In-One Image Restoration (CVPR 2025 · similarity 0.89)
- DGSolver: Diffusion Generalist Solver with Universal Posterior Sampling for Image Restoration (NeurIPS 2025 · similarity 0.88)
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.
