Comparative analysis of facial recognition models using video for real time attendance monitoring system

Attendance reporting is one of the standard processes across the world in academic institutions. The key purpose is to encourage consistency in attending school which in turn improves the learning process for a student. The manual attendance system is widely used in the educational system which is time-consuming as well as laborious. The main concept behind the automatic attendance system is to apply facial recognition effortlessly compared to other biometric systems. Following three methods i.e. Histogram Oriented Gradients (HOG), Viola-Jones (Haar Cascade), and Convolution Neural Network (CNN) are analyzed based on face detection accuracy. The Viola-Jones method delivered high accuracy amongst all. For real-time attendance systems, Viola-Jones and CNN algorithms are utilized for face detection and recognition purposes respectively. A benefit of the recommended system is to overcome hurdles like moderately detectable faces, objectionable light conditions, and alignments. The proposed system achieved 94.6% accuracy on a real-time database.