ATTENDANCE SYSTEM WITH FACIAL RECOGNITION USING MTCNN AND FACENET IN A CLASSROOM

Abstract

Attendance fraud has been a widespread issue in the attendance system. At the same time, attendance marking is essential for session performance evaluation. An attendance system is usually applied in a classroom. The traditional attendance system using paper and pen is susceptible to attendance fraud and human error. Several technologies have already been introduced to overcome the traditional attendance system issues. However, some issues are still there. Therefore, a facial recognition attendance system will give a better attendance system and reduce the drawback's possibilities. The facial recognition attendance system introduced implements several techniques. MTCNN to localize faces in an image or video frame, FaceNet to extract face features into 128-dimension face embedding, and Support Vector Machine as a classifier to the face embeddings. A web application for an attendance information system is provided for lecturers and students. Students can check their attendance status. The lecturer can load the live video of the class session and check their class attendance report. By implementing this, attendance fraud, such as marking unattended students, possibility is decreased and increases the quality and trust of the attendance data.