Acne Detection with Deep Neural Networks

In this paper, a baseline for automated acne detection with deep neural networks is presented. The goal is to overcome the poor performance of traditional acne detection methods. Acne vulgaris is a disease of sebaceous gland usually presented on the face. Nowadays, dermatologists diagnose acne by counting the number of pimples, which is time consuming. We directly compared the effectiveness of Faster-RCNN and R-FCN models. We used mean average precision (mAP) to evaluate the performance. The results confirm that acne detection with deep learning is indeed promising. The data set was taken from Pan Rajdhevee Group Public Co., Ltd. Our proposed model achieved the mean average precision of 28.3%. This is not only more accurate, but also faster than that of traditional image-processing methods.