Activating More Pixels in Image Super-Resolution Transformer

By Xiangyu Chen, Jiantao Zhou, Chao Dong. Computer Vision and Pattern Recognition, 2022

1,084 citations recorded by the source corpus.

Transformer-based methods have shown impressive performance in low-level vision tasks, such as image super-resolution. 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 reconstruction, we propose a novel Hybrid Attention Transformer (HAT). It combines both channel attention and window-based self-attention schemes, thus making use of their complementary advantages of being able to utilize global statistics and strong local fitting capability. 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 exploit the potential of the model for further improvement. Extensive experiments show the effectiveness of the proposed modules, and we further scale up the model to demonstrate that the performance of this task can be greatly improved. Our overall method significantly outperforms the state-of-the-art methods by more than 1dB.

Subfields: Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/2bbffec6395d3a8f81f5fe9137f7e078bd0f0336

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

Citations in the corpus

SeeSR: Towards Semantics-Aware Real-World Image Super-Resolution (2023, neutral): “Traditionally, researchers investigate the ISR problem by assuming simple and known image degradations ( e.g ., bicubic downsampling), and developed many successful models [7, 8, 14, 33, 41, 53, 74, 75, 79].”
MambaIR: A Simple Baseline for Image Restoration with State-Space Model (2024, methodology): “In accordance with previous works [10, 41, 78], we perform data augmentation by applying horizontal flips and random rotations of 90 [42].”
DiffBIR: Towards Blind Image Restoration with Generative Diffusion Prior (2023, background): “They have successfully promoted a vast number of excellent restoration algorithms [14; 65; 36; 7; 58; 63; 8], but are born to have limited generalization ability.”
Efficient and Explicit Modelling of Image Hierarchies for Image Restoration (2023, methodology): “On the other hand, it is shown in recent works [6,23] that even a slight increase in window size can lead to better image restoration.”
TTST: A Top-k Token Selective Transformer for Remote Sensing Image Super-Resolution (2024, methodology): “2) Window-Based Self-Attention: WSA is capable of capturing long-range dependencies, which has become an empirical operation in most existing models [11], [27].”