Integrating State-of-the-Art Face Recognition and Anti-Spoofing Techniques into Enterprise Information Systems

Satyam Mishra, Bich Thuy Nguyen Thi, Cong-Doan Truong

Abstract: Face Recognition Technology and Face Anti-Spoofing became a necessity during the Covid 19 pandemic, Monkeypox Virus etc. In the current era, the use of contactless technology has become crucial and highly beneficial for individuals. Vietnam is experiencing a significant digital transformation across various sectors including culture, education, tourism, finance, industry, and entertainment. However, most Enterprise Information System institutions in Vietnam lack facial recognition. To address this issue, we have undertaken research to devise a secure anti-spoofing method and determine an effective approach for face recognition processing. Our aim is to develop a comprehensive solution that can be implemented to establish a complete system that we researched and assessed at each stage. To construct a Facial Recognition application, we implemented a Convolutional Neural Network (CNN) as a core to recognize faces in real-time. To identify whether the faces are genuine or counterfeit, we utilized Landmark68 during the anti-spoofing phase. We applied our findings and developed an application AILib during the Covid-19 outbreak, when it was challenging for people to physically visit and login with their IDs at the counter. People can now login without physically being there by logging in using their faces on the AILib. According to the findings of our research, the system functions satisfactorily, with an ideal level of accuracy of 98.42%. Furthermore, we discovered that the optimal threshold value for identifying Asian faces in our face recognition test was determined to be 0.4, while varying threshold values were determined for different face types. For anti-spoofing, during the facial anti-spoofing test, the best threshold value for left, right and front was d < −50, d < −150 and d > −50 respectively, and the right value is d > −50 and in comparison, with state-of-the-art methods is pretty good. The program has a high level of practicality, significantly advancing the groundbreaking application of artificial intelligence to enhance people’s quality of life and safety.