Using CNN to Detect the Use of Mobile Phone in Examination Rooms

Reuben Moyo, Stanley Ndebvu, Michael Zimba, Jimmy Mbelwa

Keywords: Object Detection, Exam Integrity, CNN, Mobile Phone, Examination

Abstract: Exams play a crucial role in the learning process, and academic institutions invest significant resources to ensure their integrity by preventing cheating by students or facilitators. Unfortunately, cheating has become rampant in exam environments, compromising their integrity. The traditional method of relying on invigilators to monitor every student is impractical and ineffective. It is necessary to record exam sessions to monitor students for suspicious activities to address this challenge. However, these recordings are often too lengthy for invigilators to analyse effectively, and fatigue may cause them to miss significant details. To expand the coverage, invigilators can use fixed overhead or wearable cameras. This paper introduces a model that uses automation to analyse videos and efficiently and effectively detect the use of mobile phones as prohibited materials during exams. We used Convolutional Neural Networks (CNN) object detection techniques to identify mobile phones. The experimental results show that model achieved a 98.9% accuracy, a recall of 0.795, an F-measure of 0.697, and an average precision of 0.783. This detection system is essential in preventing cheating and promoting academic integrity, fairness, and quality education for institutions.

Submission Category: Machine learning algorithms

Submission Number: 36