Reparameterizable large kernel attention networks for infrared image super-resolution

TL;DR

This paper introduces a new network for improving infrared image quality while maintaining fast processing speeds.

Contribution

A novel Large Kernel Reparameterization Attention mechanism is proposed for infrared image super-resolution.

Findings

The proposed method improves average PSNR by 0.0008 dB on a self-constructed infrared dataset.

The network achieves 4× super-resolution on 320×180 images in 37ms on the RK3588 Neural Processing Unit.

Abstract

To address the challenge of balancing reconstruction performance and inference speed in the existing infrared image super-resolution algorithms, this paper introduces a novel Large Kernel Reparameterization Attention mechanism. Based on this, we propose the reparameterizable large kernel attention network for infrared image super-resolution. During training, a multi-branch large kernel network is employed to fully extract information, while at inference time, it is equivalently transformed into a single-branch large kernel network, achieving a trade-off between processing performance and inference speed. Compared to state-of-the-art methods, our approach improves the average PSNR on a self-constructed infrared dataset by 0.0008 dB. Additionally, on the RK3588 Neural Processing Unit, it requires only 37ms to perform 4\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$\times$$\end{document} super-resolution on 320\documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$\times$$\end{document}180 images.

Figures