Masked Face Recognition based on Attention Mechanism and FaceX-Zoo

Recently, due to the outbreak of the COVID-19 epidemic in the world, wearing face masks has become a trend, which brings difficulties to the traditional face recognition technologies that do not actively focus on the upper part of the face. This paper proposes a novel method for masked face recognition based on attention mechanism and FaceX-Zoo (an open-source method of JD.COM). In order to make the module focus on the regions around the eyes, we integrated the CBAM (Convolutional Block Attention Module) attention mechanism into ResNet50 and MobileFaceNet network. Furthermore, the FaceX-Zoo method was used to generate masked face images to improve the module performance. Experiment results show that the proposed approach can improve the performance of masked face recognition compared with competitive approaches.