• Learning to See in the Dark - Low Light Image Enhancement

Photos captured under sub-optimal lighting conditions lose information that would otherwise be evident in optimal conditions. Such conditions can be caused by inadequate and unbalanced lighting, under-exposure, or insufficient indoor lighting. Low-light enhancement is challenging because objects in a low-light image are often illuminated irregularly. The purpose of low-light image or video enhancement is to enhance or recover the information lost in the low-light regions.

Low-light image enhancement aims to improve the visibility of images taken in low-light or nighttime conditions. Currently, most deep models are trained using synthetic low-light datasets or manually collected datasets with small sizes, which limits their generalization capability when encountering the low-light images captured in the wild. In this study, a domain adaptation framework is proposed to translate images between synthetic low-light images and real low-light images. Meanwhile, we embed a method into the proposed domain adaptation framework to generate low-light images of different brightness levels, which helps with the training process of low-light enhancement networks via data augmentation. Finally, an attention-guided U-Net is trained on the augmented dataset.

Our method achieves satisfactory and comparable results when compared to state-of-the-art methods using qualitative and quantitative metrics. There is also potential in applying our method to improve face and object detection in the dark.