This repository serves as a comprehensive survey and curated list of research papers focused on diffusion model applications in image processing. It targets researchers and practitioners in computer vision and machine learning, providing a structured overview of advancements in areas like restoration, enhancement, compression, and quality assessment. The primary benefit is a centralized, up-to-date resource for understanding the rapidly evolving landscape of diffusion models in image manipulation tasks.

How It Works

The project collates and categorizes research papers, presenting them in tables that detail models, authors, training methods, venues, and specific tasks. It covers a wide array of image processing applications, including super-resolution, inpainting, denoising, dehazing, deblurring, and more, with dedicated sections for video compression and quality assessment. The organization facilitates quick identification of relevant work and trends within diffusion-based image processing.

Quick Start & Requirements

This repository is a curated list of research papers and does not involve direct installation or execution of code. All requirements are dependent on the individual papers referenced within.

Highlighted Details

• Extensive coverage of diffusion models for image restoration, including numerous sub-tasks like super-resolution, inpainting, denoising, and dehazing.
• Summaries and tables for diffusion model applications in image and video compression, as well as image quality assessment.
• Regular updates to incorporate new research, with the latest update noted as April 25, 2024.
• Includes a comprehensive list of benchmark datasets commonly used in image processing tasks.

Maintenance & Community

The repository is maintained by Xin Li and a team of contributors from institutions including the University of Science and Technology of China (USTC) and the National University of Singapore (NUS). Updates are posted regularly. Interested parties are encouraged to raise issues or email the maintainers to include their work.

Licensing & Compatibility

The repository itself does not specify a license. The content consists of links and summaries of research papers, each with its own licensing and usage terms. Compatibility for commercial use or closed-source linking would depend on the licenses of the individual papers and their associated codebases.

Limitations & Caveats

This repository is a survey and does not provide executable code or pre-trained models. Users must refer to the individual papers for implementation details and usage. The rapid pace of research means the list may not be exhaustive at any given moment, despite regular updates.