Improving Face Recognition for Mask Wearers Using Data Augmentation of Left–Right Image Flipping and Rotation

Wearing a mask hides half of a face, making it difficult to recognize using face recognition. This raises the problem of it being impossible to identify the whereabouts of a person because their face cannot be recognized. In this paper, we aim to improve the recognition rate by using a learning-based method combining an NVIDIA pre-trained model and face images with and without masks. Furthermore, we aim to improve the recognition rate by utilizing data augmentation to increase the number of training data. Experimental results show that the recognition rate of face images improved.