🤖 AI Summary
Existing Transformer-based image restoration models suffer from narrow spatial receptive fields and weak inter-window feature interaction due to rigid local window partitioning. To address these limitations, this work proposes a hybrid attention architecture comprising three key innovations: (1) a novel cooperative modeling mechanism that jointly integrates channel-wise attention with window-based self-attention; (2) an overlapping cross-attention module explicitly designed to enhance inter-window feature communication; and (3) a same-task pretraining strategy to strengthen representation learning. Extensive experiments demonstrate state-of-the-art performance across multiple low-level vision tasks—including synthetic and real-world super-resolution, Gaussian denoising, and compression artifact removal. Our method achieves significant PSNR and SSIM improvements over prior art and delivers superior visual quality with more natural textures and sharper details.
📝 Abstract
Transformer-based methods have shown impressive performance in image restoration tasks, such as image super-resolution and denoising. However, we find that these networks can only utilize a limited spatial range of input information through attribution analysis. This implies that the potential of Transformer is still not fully exploited in existing networks. In order to activate more input pixels for better restoration, we propose a new Hybrid Attention Transformer (HAT). It combines both channel attention and window-based self-attention schemes, thus making use of their complementary advantages. Moreover, to better aggregate the cross-window information, we introduce an overlapping cross-attention module to enhance the interaction between neighboring window features. In the training stage, we additionally adopt a same-task pre-training strategy to further exploit the potential of the model for further improvement. Extensive experiments have demonstrated the effectiveness of the proposed modules. We further scale up the model to show that the performance of the SR task can be greatly improved. Besides, we extend HAT to more image restoration applications, including real-world image super-resolution, Gaussian image denoising and image compression artifacts reduction. Experiments on benchmark and real-world datasets demonstrate that our HAT achieves state-of-the-art performance both quantitatively and qualitatively. Codes and models are publicly available at https://github.com/XPixelGroup/HAT.
