A Survey on All-in-One Image Restoration: Taxonomy, Evaluation and Future Trends

Abstract: Image restoration (IR) aims to recover high-quality images from inputs degraded by various factors such as noise, blur, compression, and adverse weather. Traditional IR methods typically focus on specific types of degradation, which limits their effectiveness in real-world scenarios with complex distortions. In response to this challenge, the all-in-one image restoration (AiOIR) paradigm has recently emerged, offering a unified framework that adeptly addresses multiple degradation types. These innovative models enhance convenience and versatility by adaptively learning degradation-specific features while simultaneously leveraging shared knowledge across diverse corruptions. In this survey, we present the first comprehensive overview of AiOIR, offering a taxonomy that organizes existing methods by architecture innovations, learning strategies, and key improvements. We systematically categorize prevailing approaches and critically assess the challenges these models encounter, proposing future research directions to propel this rapidly evolving field. Our survey begins with an introduction to the foundational concepts of AiOIR models, followed by a categorization of typical scenarios. We then highlight key architectural and algorithmic advances in AiOIR, aiming to inspire continued innovation. To facilitate rigorous evaluation of existing methods, we collate and summarize established datasets, evaluation metrics, and common experimental settings. Finally, we present an objective comparison of open-sourced methods, providing valuable insights for researchers and practitioners. This paper stands as the first comprehensive and insightful review of all-in-one image restoration. A related repository is available at https://github.com/Harbinzzy/All-in-One-Image-Restoration-Survey.