In the field of computer vision, improving the quality of images taken in low-light conditions remains a significant challenge. This challenge is relevant in diverse areas, from nighttime photography to advanced medical imaging. This study introduces a novel approach to address this issue, making several significant advancements. Firstly, we propose a spatial-channel attention dual-branch network, meticulously designed to harness both spatial and channel correlations in dimly lit images. This ensures a meticulous enhancement of intricate textures and details while upholding image integrity and minimizing artifacts. Secondly, we deploy a classifier to sift through various enhanced image versions, consistently selecting the most visually compelling output. In addition, we redefined traditional loss functions to better suit low-light enhancement, mitigating common issues like over-enhancement and color distortions. Empirical evaluations on multiple datasets underscore our method’s efficacy, as it outperforms leading techniques in preserving authenticity, highlighting details, and enhancing overall visual appeal. This research promises substantial improvements in the domain of low-light image processing.