An Effective Implementation of Autonomous Attendance System using Convolution Neural Networks

Abstract

Attendance marking is a common method used by all educational institutions at all levels to keep track of students' daily presence. Previously, attendance was recorded manually. These procedures are precise and remove the possibility of enrolling false attendance, but they are time-consuming and labor-intensive for a big number of pupils. Autonomous systems based on radio frequency recognition scanning, fingerprint scanning, face recognition, and iris scanning are being developed to address the drawbacks of manual systems. Each strategy has pros and cons. Furthermore, most of these systems are limited by the requirement for one-on-one human interaction to record attendance. In this work, we developed a durable and effective attendance recording system based on a single group photograph that detects face identification and recognition algorithms to solve the limitations of existing human and autonomous attendance management systems. Using a high-definition camera mounted in a fixed position, a group of photos is collected for all of the students sitting in a classroom. Following that, using a typical approach, photos of the faces are extracted from the group photo, followed by identification using a convolution neural network acquainted in a student face database. We tested our approach using a range of group pictures and datasets. In terms of efficiency, convenience of use, and implementation, the suggested framework beats existing attendance tracking systems, according to our findings. The suggested system is a self-contained attendance system with minimal human-machine interaction, making it simple to integrate into a smart classroom.