Rethinking Theoretical Illumination for Efficient Low-Light Image Enhancement

TL;DR

This paper presents CPGA-Net+ with theoretically-based attention mechanisms for efficient low-light image enhancement, achieving superior performance on edge devices with reduced computational costs.

Contribution

Introduction of CPGA-Net+ with attention mechanisms for low-light enhancement, including lightweight and stronger variants, balancing efficiency and performance.

Findings

Lightweight version reduces computational costs by over two-thirds.

Stronger version balances local and global processing effectively.

Proposed methods outperform recent lightweight approaches.

Abstract

Enhancing low-light images remains a critical challenge in computer vision, as does designing lightweight models for edge devices that can handle the computational demands of deep learning. This article introduces an extended version of the Channel-Prior and Gamma-Estimation Network (CPGA-Net), termed CPGA-Net+, incorporating the theoretically-based Attentions for illumination in local and global processing. Additionally, we assess our approach through a theoretical analysis of the block design by introducing both an ultra-lightweight and a stronger version, following the same design principles. The lightweight version significantly reduces computational costs by over two-thirds by utilizing the local branch as an auxiliary component. Meanwhile, the stronger version achieves an impressive balance by maximizing local and global processing capabilities. Our proposed methods have been validated as effective compared to recent lightweight approaches, offering superior performance and scalable solutions with limited computational resources.