Research map / Image generation / Image restoration
BD-Diff: Generative Diffusion Model for Image Deblurring on Unknown Domains with Blur-Decoupled Learning
Junhao Cheng, Wei-Ting Chen, Xi Lu, Ming–Hsuan 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 structure extractor Qs (ablation) | UID-Diff · GoPro | MANIQA: 0.2797 -> 0.3246 | Improved |
| add structure extractor Qs (ablation) | UID-Diff · GoPro | MUSIQ: 41.22 -> 54.94 | Improved |
| add blur-transfer and reconstruction joint training (Qb, T2, T3) (ablation) | UID-Diff · GoPro | MANIQA: 0.2933 -> 0.3246 | Improved |
| add blur-transfer and reconstruction joint training (Qb, T2, T3) (ablation) | UID-Diff · GoPro | LIQE: 1.91 -> 1.97 | Improved |
| add reconstruction task T3 (ablation) | UID-Diff · GoPro | MANIQA: 0.3124 -> 0.3246 | Improved |
| add reconstruction task T3 (ablation) | UID-Diff · REDS | MUSIQ: 61.89 -> 63.91 | Improved |
| add reconstruction task T3 (ablation) | UID-Diff · RealBlur-J | MUSIQ: 24.71 -> 27.06 | Improved |
| add refined VAE instead of original SD VAE (ablation) | UID-Diff · GoPro | MUSIQ: 54.69 -> 54.94 | Improved |
| add refined VAE instead of original SD VAE (ablation) | UID-Diff · GoPro | FID: 50.37 -> 50.45 | Improved |
| use UID-Diff instead of Blur2Blur (unpaired training) | UID-Diff vs Blur2Blur · RealBlur-R | MANIQA: 0.2873 -> 0.3577 | Improved |
| use UID-Diff instead of supervised Diff-Plugin | UID-Diff vs Diff-Plugin · GoPro | MANIQA: 0.2933 -> 0.3246 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- Diff-Plugin — Diff-Plugin: Revitalizing Details for Diffusion-based Low-level Tasks (CVPR 2024) · better on 4 of 4 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
- DeblurDiff: Real-Word Image Deblurring with Generative Diffusion Models (NeurIPS 2025 · similarity 0.92)
- BlurDM: A Blur Diffusion Model for Image Deblurring (NeurIPS 2025 · similarity 0.91)
- Residual Diffusion Deblurring Model for Single Image Defocus Deblurring (AAAI 2025 · similarity 0.90)
- ID-Blau: Image Deblurring by Implicit Diffusion-based reBLurring AUgmentation (CVPR 2024 · similarity 0.89)
- Blur2Blur: Blur Conversion for Unsupervised Image Deblurring on Unknown Domains (CVPR 2024 · similarity 0.88)
- GenDeg: Diffusion-based Degradation Synthesis for Generalizable All-In-One Image Restoration (CVPR 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.
