DARTS: Double Attention Reference-based Transformer for Super-resolution

TL;DR

DARTS introduces a transformer-based approach for reference-based image super-resolution that simplifies architecture and training while achieving state-of-the-art results using attention mechanisms alone.

Contribution

The paper adapts double attention blocks for super-resolution, creating a simpler, effective transformer model that outperforms complex multi-stage architectures.

Findings

Achieves state-of-the-art PSNR/SSIM on SUN80 dataset

Simplifies super-resolution architecture using attention mechanisms

Performs competitively with fewer components

Abstract

We present DARTS, a transformer model for reference-based image super-resolution. DARTS learns joint representations of two image distributions to enhance the content of low-resolution input images through matching correspondences learned from high-resolution reference images. Current state-of-the-art techniques in reference-based image super-resolution are based on a multi-network, multi-stage architecture. In this work, we adapt the double attention block from the GAN literature, processing the two visual streams separately and combining self-attention and cross-attention blocks through a gating attention strategy. Our work demonstrates how the attention mechanism can be adapted for the particular requirements of reference-based image super-resolution, significantly simplifying the architecture and training pipeline. We show that our transformer-based model performs competitively with state-of-the-art models, while maintaining a simpler overall architecture and training process. In particular, we obtain state-of-the-art on the SUN80 dataset, with a PSNR/SSIM of 29.83 / .809. These results show that attention alone is sufficient for the RSR task, without multiple purpose-built subnetworks, knowledge distillation, or multi-stage training.