Development and Optimization of Check-in System Based on Face Recognition Technology

Abstract

Face recognition technology is widely used. By comparing the speed and accuracy of Face++ and Baidu AI face recognition, the face sign-in service was collected in the mobile phone or pad APP, which realized the mobile check-in attendance that can be used for college students or training institutions. The system was characterized by accuracy, fast and anti-cheating ability. The experimental results show that Baidu AI's face detection and recognition performance is better, its speed is more than twice faster than Face++, the recognition score is higher than Face++, and includes live detection, which can effectively prevent static picture cheats. However, Baidu AI's face detection time (about 271.2ms) was much higher than the commercial face detection algorithm (about 25ms), so the Adaboost face detection algorithm was studied, and the Adaboost algorithm based on OpenCV was transplanted to Android. The platform enables real-time detection of faces. After the system optimization, the face detection time is about 10ms, which effectively improves the face detection speed.