A Transformer-Based Network for Low-Light Image Enhancement

Abstract

Challenging lighting conditions in the real world (low light, underexposure, and overexposure) not only create an unpleasant visual appearance, but also pollute computer vision tasks. we proposed a transformer-based network consists of global branch and local branch which is based on ISP theory to enhance low-light image quality. The key component in our network is the Window-based Self-Attention Block (WSAB) which captures non-local self-similarity and long-range dependencies. •Extensive experiments and ablation study demonstrate prove the efficiency of each part and demonstrate the superior performance of our proposed transformer-based network over SOTA methods.