Repository logo
 

Smart attendance monitoring system using computer vision.

Thumbnail Image

Date

2019

Journal Title

Journal ISSN

Volume Title

Publisher

Abstract

Monitoring of student’s attendance remains the fundamental and vital part of any educational institution. The attendance of students to classes can have an impact on their academic performance. With the gradual increase in the number of students, it becomes a challenge for institutions to manage their attendance. The traditional attendance monitoring system requires considerable amount of time due to manual recording of names and circulation of the paper-based attendance sheet for students to sign their names. The paper-based attendance recording method and some existing automated systems such as mobile applications, Radio Frequency Identification (RFID), Bluetooth, and fingerprint attendance models are prone to fake results and time wasting. The limitations of the traditional attendance monitoring system stimulated the adoption of computer vision to stand in the gap. Student’s attendance can be monitored with biometric candidate’s systems such as iris recognition system and face recognition system. Among these, face recognition have a greater potential because of its non-intrusive nature. Although some automated attendance monitoring systems have been proposed, poor system modelling negatively affects the systems. In order to improve success of the automated systems, this research proposes the smart attendance monitoring system that uses facial recognition to monitor student’s attendance in a classroom. A time integrated model is provided to monitor student’s attendance throughout the lecture period by registering the attendance information at regular time intervals. Multi-camera system is also proposed to guarantee an accurate capturing of students. The proposed multi-camera based system is tested using a real-time database in an experimental class from the University of KwaZulu-Natal (UKZN). The results show that the proposed smart attendance monitoring System is reliable, with the average accuracy rate of 98%.

Description

Masters Degree. University of KwaZulu-Natal, Durban.

Keywords

Citation

DOI