DEVELOPMENT OF A FACE MASK DETECTION SYSTEM USING SINGLE SHORT ALGORITHM: A CASE STUDY OF ELIZADE UNIVERSITY

Abstract

This paper discusses the development of a Face Mask Detection System using a Single Short algorithm for the prevention of the
spread of COVID-19 in public places. Several works have been done in the detection of face masks; however, there is a need to
increase the detection speeds while maintaining their high accuracy for large datasets. The developed system consists of both
software and hardware components. The model of the system was developed with a Single Short algorithm with a total of Nine
Hundred and Two (902) datasets with the faces of people with and without face masks, which were collected from Elizade
University, Ilara-Mokin, Ondo State of Nigeria. The Single Short Detection MobileNetv2 Algorithm was used to develop a
predictive model and deployed on the Raspberry Pi 4 microcontroller. Percentage accuracy, F1 score, Recall, and Precision were
the performance evaluation metrics used for the work. Also, a questionnaire was distributed to fifty (50) participants, mostly
students and staff of Elizade University, Ilara-Mokin, who tested the system with and without wearing a face mask. The result of
the system‟s performance evaluation indicates an accuracy of 99.86%, an F1 score of 100%, a recall of 100%, and a precision of
100%. The developed system can be miniaturised and reproduced to make the entire system smaller and more affordable. With the availability of the system‟s prototype, the development of the system for access control in public places such as stadiums,
shopping malls, and schools is possible.