Restormer: Efficient Transformer for High-Resolution Image Restoration

TL;DR

Restormer is an efficient Transformer-based model designed for high-resolution image restoration tasks, achieving state-of-the-art results while maintaining computational feasibility.

Contribution

The paper introduces a novel efficient Transformer architecture tailored for high-resolution image restoration, addressing the quadratic complexity issue of standard Transformers.

Findings

Achieves state-of-the-art results on multiple image restoration tasks.

Effectively captures long-range pixel interactions in large images.

Maintains computational efficiency suitable for high-resolution images.

Abstract

Since convolutional neural networks (CNNs) perform well at learning generalizable image priors from large-scale data, these models have been extensively applied to image restoration and related tasks. Recently, another class of neural architectures, Transformers, have shown significant performance gains on natural language and high-level vision tasks. While the Transformer model mitigates the shortcomings of CNNs (i.e., limited receptive field and inadaptability to input content), its computational complexity grows quadratically with the spatial resolution, therefore making it infeasible to apply to most image restoration tasks involving high-resolution images. In this work, we propose an efficient Transformer model by making several key designs in the building blocks (multi-head attention and feed-forward network) such that it can capture long-range pixel interactions, while still remaining applicable to large images. Our model, named Restoration Transformer (Restormer), achieves state-of-the-art results on several image restoration tasks, including image deraining, single-image motion deblurring, defocus deblurring (single-image and dual-pixel data), and image denoising (Gaussian grayscale/color denoising, and real image denoising). The source code and pre-trained models are available at https://github.com/swz30/Restormer.