Occluded-based Facial Recognition under mask using Deep learning method

At present, under the epidemic prevention and control, face recognition is facing the problem of masks. The traditional face recognition system dataset is composed of a face without mask pictures, and when the mask wearers need to do the face recognition, they have to take off the mask, which makes the epidemic situation environment increase the burden of prevention and control. Therefore, effectively improving the efficiency of face recognition under face masking is still a difficult point in the recognition system. In this paper, we aim to conduct facial mask detection to detect the person with or without the mask. Three steps are included in our method, including facial detection, adding mask and facial detection with mask. To achieve this goal, we aim to use MTCNN and FaceNet as the main components of our method. Our method can achieve 76.42% accuracy among all the data. The results on both real and generated dataset validate the effectiveness of out method.