lixinustc/Awesome-diffusion-model-for-image-processing

Curated repository organizing 200+ diffusion-based papers across restoration tasks (super-resolution, denoising, inpainting, deblurring), compression, and quality assessment with structured tables, benchmarks, and implementation links. Leverages conditional diffusion architectures and score-based generative models adapted for low-level vision tasks, with emphasis on efficiency improvements like latent-space diffusion and single-step sampling. Covers supervised learning approaches published across CVPR, ICCV, NeurIPS with active curation through community contributions and quarterly updates.

927 stars.