SAAT: Synergistic Alternating Aggregation Transformer for Image Super-Resolution

TL;DR

SAAT introduces a novel transformer-based model that synergistically combines channel and spatial attention mechanisms to improve image super-resolution, achieving state-of-the-art performance with efficient feature utilization.

Contribution

The paper proposes SAAT, a new model that effectively integrates channel and spatial attention through innovative attention groups, enhancing feature extraction for super-resolution.

Findings

SAAT achieves competitive super-resolution results with fewer parameters.

The synergistic attention groups improve feature fusion and structural information capture.

Ablation studies confirm the effectiveness of the proposed attention mechanisms.

Abstract

Single image super-resolution is a well-known downstream task which aims to restore low-resolution images into high-resolution images. At present, models based on Transformers have shone brightly in the field of super-resolution due to their ability to capture long-term dependencies in information. However, current methods typically compute self-attention in nonoverlapping windows to save computational costs, and the standard self-attention computation only focuses on its results, thereby neglecting the useful information across channels and the rich spatial structural information generated in the intermediate process. Channel attention and spatial attention have, respectively, brought significant improvements to various downstream visual tasks in terms of extracting feature dependency and spatial structure relationships, but the synergistic relationship between channel and spatial attention has not been fully explored yet.To address these issues, we propose a novel model. Synergistic Alternating Aggregation Transformer (SAAT), which can better utilize the potential information of features. In SAAT, we introduce the Efficient Channel & Window Synergistic Attention Group (CWSAG) and the Spatial & Window Synergistic Attention Group (SWSAG). On the one hand, CWSAG combines efficient channel attention with shifted window attention, enhancing non-local feature fusion, and producing more visually appealing results. On the other hand, SWSAG leverages spatial attention to capture rich structured feature information, thereby enabling SAAT to more effectively extract structural features.Extensive experimental results and ablation studies demonstrate the effectiveness of SAAT in the field of super-resolution. SAAT achieves performance comparable to that of the state-of-the-art (SOTA) under the same quantity of parameters.