Review Learning: Advancing All-in-One Ultra-High-Definition Image Restoration Training Method

TL;DR

This paper introduces Review Learning, a new training paradigm for UHD image restoration that improves model memory and performance across multiple degradation types without prior knowledge, using sequential training and a lightweight network.

Contribution

The paper proposes Review Learning, a novel training method that enhances generalization and memory in UHD image restoration models without relying on prompts or customized networks.

Findings

Effective handling of multiple degradation types without prior knowledge.

Improved performance on 4K UHD images with a lightweight network.

Sequential training with review mechanism enhances model memory.

Abstract

All-in-one image restoration tasks are becoming increasingly important, especially for ultra-high-definition (UHD) images. Existing all-in-one UHD image restoration methods usually boost the model's performance by introducing prompt or customized dynamized networks for different degradation types. For the inference stage, it might be friendly, but in the training stage, since the model encounters multiple degraded images of different quality in an epoch, these cluttered learning objectives might be information pollution for the model. To address this problem, we propose a new training paradigm for general image restoration models, which we name \textbf{Review Learning}, which enables image restoration models to be capable enough to handle multiple types of degradation without prior knowledge and prompts. This approach begins with sequential training of an image restoration model on several degraded datasets, combined with a review mechanism that enhances the image restoration model's memory for several previous classes of degraded datasets. In addition, we design a lightweight all-purpose image restoration network that can efficiently reason about degraded images with 4K ($3840 \times 2160$) resolution on a single consumer-grade GPU.