Research map / Image generation / Fast diffusion sampling
Diffusion Models in Low-Level Vision: A Survey
Chunming He, Yuqi Shen, Chengyu Fang, Fengyang Xiao, Longxiang Tang, Yulun Zhang, Wangmeng Zuo, Zhenhua Guo, Xiu Li
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 DiffIR instead of IR-SDE for 4x super-resolution | DM-based SR · DIV2K 4x SR | PSNR: 25.90 -> 29.13 | Improved |
| use DiffIR instead of IR-SDE for 4x super-resolution | DM-based SR · DIV2K 4x SR | time per image (s): 63.9 -> 0.3 | Improved |
| use DiffIR on Urban100 instead of ResDiff | DM-based SR · Urban100 4x SR | PSNR: 27.43 -> 26.05 | Hurt |
| use DiffIR instead of IR-SDE for motion deblurring | DM-based deblurring · GoPro motion deblurring | PSNR: 30.70 -> 33.20 | Improved |
| use DiffIR instead of IR-SDE for motion deblurring | DM-based deblurring · GoPro motion deblurring | time per image (s): 4.3 -> 0.436 | Improved |
| use Tiramisu instead of DDNM for zero-shot inpainting | DM-based inpainting · ImageNet 1K (Half mask) | LPIPS: 0.408 -> 0.286 | Improved |
| use Tiramisu instead of DDRM for zero-shot inpainting | DM-based inpainting · ImageNet 1K (Half mask) | time per image (s): 4.9 -> 186.5 | Hurt |
| use GSAD with gt mean strategy instead of PyDiff | DM-based low-light enhancement · LOLv2-Real | PSNR: 24.01 -> 28.82 | Improved |
| transfer GSAD from LOLv2-Syn to LOLv1 | GSAD · LOLv2-Syn to LOLv1 | PSNR: 28.67 -> 20.48 | Hurt |
Closest papers by meaning
- Diff-2-in-1: Bridging Generation and Dense Perception with Diffusion Models (ICLR 2025 · similarity 0.92)
- Aligning Generative Denoising with Discriminative Objectives Unleashes Diffusion for Visual Perception (ICLR 2025 · similarity 0.91)
- Warm Diffusion: Recipe for Blur-Noise Mixture Diffusion Models (ICLR 2025 · similarity 0.90)
- Neural Diffusion Models (ICML 2024 · similarity 0.90)
- Diffusion Priors for Variational Likelihood Estimation and Image Denoising (NeurIPS 2024 · similarity 0.90)
- Simpler Diffusion: 1.5 FID on ImageNet512 with Pixel-space Diffusion (CVPR 2025 · similarity 0.90)
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.
