wyhuai/DDNM

Health v2 · maintenance only

Flags: no_releases

Adoption not part of the score

1349 stars · 107 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-30, confidence not recorded

DDNM is a Python research codebase implementing the Denoising Diffusion Null-Space Model for zero-shot image restoration, published as an ICLR 2023 Oral paper. It restores images from degradations like super-resolution, denoising, colorization, inpainting, deblurring, and compressed sensing without any task-specific training or optimization.

Use cases

• restore old photos with diffusion models
• zero-shot image super-resolution
• colorize black and white images
• inpaint missing regions in images
• deblur blurry photos
• denoise images without training a model
• solve compressed sensing image reconstruction

When to choose

• you need image restoration without training a task-specific model
• you want a research-grade diffusion-based restoration baseline
• you need to handle arbitrary image sizes or multiple degradation types with one framework

When to avoid

• you need a fast, production-ready image enhancement tool
• you lack a GPU or cannot download large pre-trained diffusion checkpoints
• you need real-time or mobile image restoration

Facets

library · maturity stable

image-processing machine-learning deep-learning computer-vision image-processing deep-learning artificial-intelligence python diffusion-models zero-shot image-restoration super-resolution inpainting colorization deblurring denoising research-code iclr2023 gpu linux

Member repositories

For agents

markdown · JSON · MCP: product_card(name="wyhuai/DDNM")