AI Powered Malpractice Detection System for Exam

Abstract

This project introduces an automated Examination Malpractice Detection System that utilizes YOLOv8 for real-time object detection and MediaPipe for human pose estimation. By processing live video streams through a Flask-based backend, the system identifies prohibited items and suspicious behavioral patterns such as excessive head movement or unauthorized communication. To ensure high reliability, it employs a 60% confidence threshold and a decision-fusion layer that reduces false positives by analyzing sequences of motion. The architecture features a seamless administrative response layer that captures time-stamped evidence and compiles infractions into a secure zip archive. Upon session completion, the system automatically transmits a comprehensive summary and evidence file to examiners via an integrated email notification service, ensuring a transparent and objective disciplinary review process.

Introduction

The text describes an AI-based Examination Malpractice Detection System designed to improve fairness and integrity in both physical and online exams by replacing traditional human invigilation with automated monitoring.

It explains that manual proctoring is limited by fatigue, bias, and inability to monitor large groups effectively, leading to increasing exam malpractice. To solve this, the proposed system uses Artificial Intelligence, computer vision, and deep learning to monitor students in real time.

The system combines object detection (using YOLO) to identify prohibited items like phones or smartwatches and human pose estimation (like OpenPose or MediaPipe) to track suspicious body movements such as unusual head turns or hidden communication. It also includes facial recognition and biometric verification to ensure each detected incident is linked to the correct student, creating a secure evidence trail.

A key improvement is its ability to reduce false alarms by analyzing behavior over time instead of single frames, helping distinguish normal actions (like stretching) from cheating behavior. The system is designed to be scalable, low-latency, and applicable to both classrooms and remote exams.

The literature review shows that previous research has explored AI-based proctoring using CNNs, RNNs, YOLO, pose estimation, and behavioral analysis. While these systems improve detection accuracy, challenges remain such as dataset limitations, false positives, ethical concerns, and difficulty linking cheating events to student identities.

The proposed system addresses these gaps by integrating multiple technologies into a unified framework that performs real-time monitoring, behavior analysis, object detection, and identity tracking, creating a more reliable and automated exam supervision solution.

Conclusion

The development of the proposed Examination Malpractice Detection System marks a significant advancement in the integration of Artificial Intelligence within educational assessment frameworks. By synthesizing high-speed object detection through YOLOv8 with the nuanced behavioral analysis of Human Pose Estimation, this research addresses the critical limitations of traditional human invigilation, such as subjectivity, scalability, and the potential for oversight. The implementation of a multi-layered architecture ensures that the system can operate in real-time, providing an objective and persistent monitoring solution that upholds academic integrity without the biases inherent in manual supervision. The results of this study demonstrate that the synergy between deep learning models and a robust XGBoost-based decision-fusion layer significantly reduces the frequency of false positives while maintaining a high detection rate for prohibited activities. Furthermore, the seamless integration of a facial recognition sub-routine and institutional bio-data ensures a level of accountability previously unattainable, linking every flagged infraction directly to a verified student identity. This digital evidence trail not only acts as a powerful deterrent against dishonest practices but also provides administrative bodies with a transparent and indisputable basis for disciplinary actions. Ultimately, this research underscores the transformative potential of AI in fostering a fair and secure academic environment. While the system provides a comprehensive technical solution for current malpractice challenges, it also establishes a scalable foundation for future enhancements, such as the incorporation of audio-visual fusion and edge computing for even lower latency. As educational institutions continue to evolve toward more flexible and digital-first testing models, the deployment of such intelligent surveillance systems will be essential in preserving the value of academic qualifications and ensuring that merit remains the sole metric of student success.

References

[1] Adeyemi, J. O., Ogunlere, S. O., & Akwaronwu, B. G. (2025). Real-time detection of examination malpractices using convolutional neural networks and video surveillance: A systematic review with meta-analysis. British Journal of Computer, Networking and Information Technology, 8(2), 15-50. [2] Ahmed, M. K., & Mohammed, G. J. (2025). Advancements in automated cheating detection systems for online and in-person examinations: A comprehensive review of methods, technologies, and effectiveness. International Journal of Mechatronics, Robotics, and Artificial Intelligence, 1(2), 124-142. [3] Ayonote, W. E. (2025). Assessment of ICT usage in the control of examination malpractice in the National Open University of Nigeria (NOUN). International Journal of Education Humanities and Social Science, 8(4). [4] Ekine-Pakaye, A. C., & Agbo, O. C. (2025). Enhanced examination malpractice detection model using XGBoost and Decision Tree techniques. Innovative Journal of Science and Technology Research, 12(2), 80-87. [5] Kalagbor-Gbeke, I. (2026). Integration of artificial intelligence tools in the management of examination malpractice in public tertiary institutions in Rivers State, Nigeria. International Journal of Educational Management, Rivers State University, 2(2), 154-167. [6] Noma-Osaghae, E., David, U. J., Mommoh, J. S., Adetunji, O. J., & Isaac, O. A. (2025). Development of an examination hall malpractice detection system using pose estimation and machine learning. Mansoura Engineering Journal, 50(4), Article 13. [7] Ofor-Douglas, S. (2026). Implementation of artificial intelligence for minimization of examination malpractice in the Nigerian university educational system. International Journal of Educational Management, Rivers State University, 2(1), 332-350. [8] Shruthi, S. V., & Chethan, H. K. (2025). A survey of machine learning techniques for intelligent proctoring systems. Journal of Design (Zheji Xuebao), 11(5), 290-302.

Copyright

Copyright © 2026 K Simeon Victor, Ehtesham Azam Syed, Shaik Rizwan Ullah Ansari, Dr. Shahan Tanveer. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.