An approach towards development of automated attendance system using face detection and recognition

Conventionally, the process of taking attendance of students in a classroom is quite a laborious task, wherein either the teacher has to call out names of each individual student, or the student has to sign an attendance sheet. In recent times, due to the Covid-19 pandemic, special importance has been laid on facial recognition techniques, which are contact-free (unlike fingerprint scanners), and are in accordance with social distancing norms. In this paper, a software system automating the attendance-taking scheme is presented. This software integrates face detection, image processing and face recognition approaches to come up with a consolidated attendance system capable of overcoming the disadvantages of manual attendance. In the system, an end user has to first log in and subsequently, an IP camera (which is to be installed in the classroom) gets turned on, and the camera starts taking photographs of the classroom. The user can also manually upload images into the system, in case calculation of attendance is not immediately required. The Histogram of Oriented Gradients (HOG) approach is employed for the face detection mechanism in the proposed system. After a comparative performance analysis of different facial recognition techniques, the Local Binary Patterns Histograms (LBPH) method is chosen as the facial recognition procedure for the system. Once all the individual students have been recognised by comparison with the model built from the extracted faces, the final results are sorted according to date (similar to taking attendance of a class on a particular day) and stored in the database.