Restormer: Efficient Transformer for High-Resolution Image Restoration
This work addresses a key bottleneck in image restoration for researchers and practitioners by enabling efficient Transformer use on high-resolution images, though it is incremental as it builds on existing Transformer and CNN methods.
The authors tackled the problem of applying Transformer models to high-resolution image restoration by proposing Restormer, an efficient Transformer that captures long-range pixel interactions while remaining computationally feasible for large images, achieving state-of-the-art results on tasks like image deraining, deblurring, and denoising.
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.