Abstract
Low-light image enhancement is a crucial preprocessing task for some complexvision tasks. Target detection, image segmentation, and image recognitionoutcomes are all directly impacted by the impact of image enhancement. However,the majority of the currently used image enhancement techniques do not producesatisfactory outcomes, and these enhanced networks have relatively weakrobustness. We suggest an improved network called BrightenNet that uses U-Netas its primary structure and incorporates a number of different attentionmechanisms as a solution to this issue. In a specific application, we employthe network as the generator and LSGAN as the training framework to achievebetter enhancement results. We demonstrate the validity of the proposed networkBrightenNet in the experiments that follow in this paper. The results itproduced can both preserve image details and conform to human vision standards.