A UV Spot Detection Method Based on Facial Skin Image

Skin health and medical aesthetics have always been widely discussed, with ultraviolet(UV) spot being a common skin concern. Currently, the detection of UV spot primarily relies on specialized image acquisition devices. These devices are complex to operate and have strict requirements for the external environment, limiting their widespread application in practical scenarios. Therefore, we propose a UV spot detection method based on generative adversarial networks. In this work, We construct a paired dataset comprising UV images of facial skin and corresponding UV spot images. We employ a feature extraction module based on dilated convolutions to expand the receptive field. Furthermore, we introduce Res2net between the encoder and decoder and integrate the Shuffle Attention to enhance the network’s feature learning capability. The Experimental results indicate that our method outperforms existing image translation methods in UV spot detection tasks.