Online Exam Fraud Detection Using Machine Learning and Computer Vision: Real-Time Monitoring and Risk Assessment
P U Anitha,
K Manaswini,
P Rajendhar,
Md.Yakub Pasha,
P Shashank
This paper presents the design and development of an Online Exam Fraud Detection System using
machine learning and computer vision techniques to ensure fairness and integrity in digital examinations.
The system is developed using Python and Streamlit, integrating YOLO-based object detection and
MediaPipe facial landmark tracking to monitor student behavior in real time. It detects suspicious
activities such as multiple faces, unauthorized devices, and abnormal head or eye movements. The system
includes modules for authentication, fraud detection, and admin monitoring with real-time alerts and
evidence storage. Experimental evaluation demonstrates that the proposed system improves accuracy,
reduces manual invigilation effort, and provides a scalable solution for secure online assessments.
