This repository provides a comprehensive survey of "All-in-One" (AiO) image restoration techniques, categorizing methods, evaluating their performance, and outlining future research directions. It serves as a valuable resource for researchers and practitioners in computer vision and image processing seeking to understand and advance the field of unified image restoration.

How It Works

The survey systematically reviews a wide range of AiO image restoration methods, including those based on transformers, diffusion models, and other advanced architectures. It analyzes their approaches to handling multiple degradation types simultaneously, often through techniques like prompt learning, degradation-aware feature extraction, and multi-task learning, aiming for generalizable and robust performance.

Quick Start & Requirements

This repository is a survey and does not require installation or execution of code. It provides links to the survey paper and a curated list of relevant research papers with their associated venues and code availability.

Highlighted Details

• Extensive compilation of over 100 AiO image restoration papers, categorized by year and approach.
• Detailed performance comparisons across multiple challenging datasets (e.g., SOTS, Rain100L, BSD68, GoPro, LOL, CDD-11, FoundIR).
• Includes a taxonomy of AiO restoration methods, highlighting key architectural trends and techniques.

Maintenance & Community

The survey is actively maintained, with recent updates reflecting advancements in the field. Contributions are welcomed via pull requests or email to the provided address.

Licensing & Compatibility

The repository itself is likely under a permissive license (e.g., MIT, Apache 2.0) as it primarily hosts survey information. However, the linked research papers will have their own respective licenses.

Limitations & Caveats

As a survey, this repository does not offer executable code for the methods discussed. The performance metrics presented are based on the original papers and may vary depending on implementation and evaluation setups.