Real-Time Monitoring System Based for Detecting Cheating in Exams Using Machine Learning and Deep Learning Approaches

Abstract

In the rapidly evolving digital era, technology is crucial in various aspects of life, including education. This research aims to detect cheating during paper-based exams using real-time motion detection technology to uphold exam integrity and honesty. The study proposes a real-time motion-based cheating detection system for manual exams, utilizing deep learning and machine learning techniques to improve detection accuracy and speed. The method involves collecting data from exam video recordings, extracting features with Convolutional Neural Networks (CNN), and classifying them using Support Vector Machine (SVM) and YOLOv5 algorithms. The findings demonstrate that the YOLOv5 method is highly effective in identifying suspicious behaviors, achieving a precision of 0.90 and a recall of 0.88, resulting in an overall accuracy of 90%. The system can detect suspicious head and body movements by focusing on exam papers and student faces. Consequently, this research significantly enhances the integrity and honesty of educational evaluation systems. Moreover, integrating this technology ensures that anomalies are quickly identified and addressed, providing a fair testing environment for all students. Future research may involve refining the system's ability to differentiate between regular and suspicious movements more accurately and incorporating other modalities, such as audio cues, for a more comprehensive monitoring approach. This study underscores the potential of advanced technological solutions in maintaining academic integrity and fairness in educational settings, suggesting that continued innovation in this area is critical for the future of secure educational assessments.

Access to Document

Other files and links

• Link to publication in Scopus
• Open Access Version Available

Fingerprint