Development of an Automated Eye Gaze Tracking System for Detecting Malpractice in Online Assessment
Abstract
This study presents the development of an automated eye gaze monitoring system
designed to detect possible malpractice during online examinations. It addresses
the increasing reliance on digital learning and the challenge of maintaining
academic integrity in remote environments where traditional supervision is
limited. The system uses computer vision techniques with Python, incorporating
libraries like OpenCV and MediaPipe to track eye gaze direction in real-time. A
rule-based method flags suspicious behaviors such as prolonged off-screen glances
or rapid eye movements that may suggest cheating. The system includes a userfriendly
web interface developed with HTML, CSS, and JavaScript, backed by a
FastAPI server for efficient processing. It was tested on 10 students across
academic levels (100 to 400), achieving an average accuracy of over 85% in gaze
and malpractice detection, as detailed in Chapter 4. With low computational
demand and a privacy-focused approach that avoids facial recognition or audio
monitoring, the system offers a scalable solution for institutions and supports
future enhancements such as machine learning integration or LMS compatibility.