Research map / Image generation / Image restoration
DiffIR: Efficient Diffusion Model for Image Restoration
Bin Xia, Yulun Zhang, Shiyin Wang, Yitong Wang, Xinglong Wu, Yapeng Tian, Wenming Yang, Luc Van Gool
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 |
|---|---|---|---|
| use diffusion model to estimate the compact IR prior (vs no DM) (ablation) | DiffIR_S2 · CelebA-HQ inpainting | FID: 5.6782 -> 5.1440 | Improved |
| joint optimization of denoising network, CPEN_S2 and DIRformer instead of traditional DM optimization (ablation) | DiffIR_S2 · CelebA-HQ inpainting | FID: 5.9766 -> 5.1440 | Improved |
| traditional DM optimization with DM compared with no DM (ablation) | DiffIR_S2 · CelebA-HQ inpainting | FID: 5.6782 -> 5.9766 | Hurt |
| insert variance noise in reverse diffusion process (ablation) | DiffIR_S2 · CelebA-HQ inpainting | FID: 5.1440 -> 5.1937 | Hurt |
| use L1-style L_diff instead of L2 for IPR estimation (ablation) | DiffIR_S2 · CelebA-HQ inpainting | FID: 5.1837 -> 5.1440 | Improved |
| use L_diff instead of KL loss for IPR estimation (ablation) | DiffIR_S2 · CelebA-HQ inpainting | FID: 5.2365 -> 5.1440 | Improved |
| increase diffusion iterations up to 3-4 (ablation) | DiffIR_S2 · CelebA-HQ inpainting | performance | Improved |
| increase diffusion iterations beyond 4 (ablation) | DiffIR_S2 · CelebA-HQ inpainting | performance | No clear effect |
| estimate a compact IPR with DM instead of whole images, versus RePaint | inpainting · CelebA-HQ wide masks | FID: 5.4881 -> 5.1440 with 4.3% parameters and 0.1% compute | Improved |
| use DiffIR_S2 instead of Restormer for deblurring | motion deblurring · GoPro | PSNR: 32.92 -> 33.20 | Improved |
| use ground-truth-guided stage-1 prior (DiffIR_S1) vs estimated prior (S2) (ablation) | DiffIR · CelebA-HQ inpainting | FID: 4.8045 -> 5.1440 | Mixed |
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.
- Restormer — Restormer: Efficient Transformer for High-Resolution Image Restoration (CVPR 2022) · better on 2 of 2 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
- A Modular Conditional Diffusion Framework for Image Reconstruction (NeurIPS 2024 · similarity 0.89)
- Residual Denoising Diffusion Models (CVPR 2024 · similarity 0.89)
- Blind Image Restoration via Fast Diffusion Inversion (NeurIPS 2024 · similarity 0.89)
- Reconciling Stochastic and Deterministic Strategies for Zero-shot Image Restoration using Diffusion Model in Dual (CVPR 2025 · similarity 0.89)
- IRBridge: Solving Image Restoration Bridge with Pre-trained Generative Diffusion Models (ICML 2025 · similarity 0.89)
- Deep Equilibrium Diffusion Restoration with Parallel Sampling (CVPR 2024 · 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.
