UnSCAR: Universal, Scalable, Controllable, and Adaptable Image Restoration

2026-03-08

AI Summary

This work addresses the limitations of existing general-purpose image restoration methods, which suffer from training instability, model bloat, and performance degradation when handling diverse degradation types, hindering their scalability to complex real-world scenarios. The authors propose a unified inference framework that, for the first time, identifies catastrophic task forgetting in multi-degradation joint learning. To mitigate this, they introduce a multi-branch mixture-of-experts architecture that decouples restoration knowledge into specialized, adaptable, and controllable expert modules. The proposed method enables stable training across more than 16 degradation types and achieves state-of-the-art performance on both seen and unseen degradation domains, significantly enhancing the scalability, controllability, and generalization capability of universal image restoration.

Technology Category

• Computer Vision: Multi-modal Vision
• Machine Learning: Multi-instance/Multi-view Learning
• Search and Optimization: Learning to Search

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs
• Responsible Web: Machine-in-the-loop, human agency and autonomy

Abstract

Universal image restoration aims to recover clean images from arbitrary real-world degradations using a single inference model. Despite significant progress, existing all-in-one restoration networks do not scale to multiple degradations. As the number of degradations increases, training becomes unstable, models grow excessively large, and performance drops across both seen and unseen domains. In this work, we show that scaling universal restoration is fundamentally limited by interference across degradations during joint learning, leading to catastrophic task forgetting. To address this challenge, we introduce a unified inference pipeline with a multi-branch mixture-of-experts architecture that decomposes restoration knowledge across specialized task-adaptable experts. Our approach enables scalable learning (over sixteen degradations), adapts and generalizes robustly to unseen domains, and supports user-controllable restoration across degradations. Beyond achieving superior performance across benchmarks, this work establishes a new design paradigm for scalable and controllable universal image restoration.

Problem

• universal image restoration
• scalability
• degradation interference
• task forgetting
• multi-degradation

Innovation

• universal image restoration
• mixture-of-experts
• scalable learning
• controllable restoration
• catastrophic forgetting