A Practical Solution Towards Development of Real-Time Face Attendance System

Face recognition is an interesting and familiar research area in biometric with many more applications. Face recognition-based attendance system is one of the effective applications in the area of biometrics. Marking the attendance manually is a difficult task for a large corpus, identifying unconnected persons and also poses many challenges. The proposed work is an effective solution to overcome those problems by the way of developing a real-time face attendance system. It subsequently combines the process of classification and then matching for the person identification. Through the experimentation, minimal sample requirement for each person and suitable learning algorithm is identified for the proposed work. The respective samples are found out by an effective clustering algorithm. Also, proposed a novel method to compute the matching threshold. Comparative study is done with existing state of the art result face attendance systems. The proposed system is evaluated with 200 students of age ranging from 21–25 years for a semester long and noticed that the system outperforms with identifying unconnected people with 100% True Acceptance Rate with a matching threshold of 0.43. Finally, the proposed system maintains entire tasks of attendance management, generates reports and notifies to the admin in a dashboard.