What is awesome-diffusion-model-for-image-processing?

This repository is a curated reading list for anyone interested in how "diffusion models", the same AI technology behind popular image generators like Midjourney or DALL-E, are being used to fix, improve, and compress existing photos and videos. Instead of creating art from scratch, these tools take damaged or low-quality images and make them clear, sharp, and usable. Think of diffusion models as a smart version of "fill in the blanks." They start with visual noise and gradually refine it step-by-step until a clean picture emerges. The research papers collected here apply that concept to everyday visual problems: making a tiny, pixelated photo large and crisp (super-resolution), removing watermarks or fixing old photos (image restoration), cleaning up dark or hazy shots (enhancement), and even shrinking image file sizes efficiently (compression) or judging image quality automatically. The primary audience is researchers, machine learning students, and engineers working on computer vision products. For example, if a startup is building an app to restore old family photographs or a company wants to enhance low-resolution medical scans like MRIs, the developers could use this list to find the exact academic papers and open-source code needed to build that feature. It serves as a helpful map to a fast-moving niche in AI research. At a high level, the project is simply an organized set of tables. It categorizes dozens of academic papers by their specific task, lists the authors, notes where the research was published, and links to both the paper itself and any companion code repositories. It also highlights a larger "survey" paper written by the maintainers that summarizes the entire field. The project is maintained by a team of researchers from universities and tech companies. It is actively updated, with the curators regularly adding newly published research and inviting other developers to submit their own papers to keep the collection current.