Face Recognition with and Without Mask Using Machine Learning Approach

Abstract

The global spread of COVID-19 has had an immediate effect on our everyday lives by disrupting international commerce and transportation. Mask use is highly recommended to prevent the spread of illness. Consequently, covering one's face with a mask has become mandatory. When everyone is hiding their identities behind masks, facial recognition software can't pick out a single face. Many businesses and organizations providing public services mandate that customers and the employees should wear protective masks. Therefore, identifying masks and who is the person behind the mask are essential in serving the global community. To solve this problem, we provide a face recognition method that is able to tell the difference between masked and unmasked faces and can also recognize the person under the mask. In this research, we propose combining MobileNetV2, VGG19 with the HOG method for fast and accurate recognition. The proposed method is tested extensively and shows promising results (94% accuracy in training and testing).