FACE RECOGNITION WITH/WITHOUT MASK

Abstract

Face recognition is a research area of computer vision. The facial recognition system can recognize faces, we propose to combine recognition of both mask-wearing and non[1]masking people. In this paper, we present the MTCNN technique to help identify and align faces. ArcFace technique to recognize facial features and extracts facial features. In addition, we propose to MobileNet which can reduce several million parameters but still maintain stable accuracy combined with ResNet to increase accuracy when recognizing faces with/without wearing a mask. The results of the problem have been tested on the RMFRD database, giving results with an accuracy of over 72%