Empirical Study on Web-Based Facial Recognition Services

Face recognition as an identity authentication has been widely used in commodity mobile systems. However due to increasing attacks against face recognition algorithms, mobile systems users begin to concern about the security and reliability of face recognition modules in their mobile devices. In China, many commercial devices manufacturers prefer to adapt web-based API to their devices as the user authentication method. In this paper, we focus on accuracy and reliability performance in both Baidu and Face++ face comparison API. To make a comparison, we also tested on Google facenet as the baseline. The result shows that liveness detection method has not been widely applied in web based API. Those without liveness detection will be vulnerable to presentation attack. Depends on the security level, a certain mount of presentation attack could be prevented by current liveness detection method.