A Secure Face Anti-spoofing Approach Using Deep Learning

Abstract

Face recognition is an attractive field for researchers since the past years and it is going to be the most practical method in identity recognition in the future. But even modern face detection systems have the problem of spoofing and it is safe to say it is the most serious obstacle in the way of becoming practical in more secure fields of identity recognition. In this paper, firstly, we discuss object detection systems in particular face detection and their problems such as using the person's picture in mobile phones or any other screen-based devices instead of a real person's face and also using masks to make faces like someone's face or other spoofing goals. Then, we introduce our method to prevent these attacks to have a high-secure and reliable face recognition system. Finally, some illustrations will be provided to demonstrate the practical and economic benefits of the presented approach.