A Survey of Masked Face Recognition Methods and Corpora/Data

Public safety in modern cities is very relevant. A lot of crime investigation cases can be solved, and a lot of crime incidents can be prevented with appliance of face recognition methods. Urban areas are equipped with security cameras, and in many cities, face recognition methods are applied. In addition, many urban services such as ATMs and payment points in subway use facial recognition. However, these systems are not efficient enough for occluded face analysis. This paper analyzes main methods for masked face recognition by images. Moreover, this paper provides a comparison of images datasets for person recognition systems. The findings demonstrated that the most effective methods of face recognition are those that are based on convolutional neural networks and the mask area feature extraction. The findings support the identifying of requirements for perspective masked person recognition systems.