AI-Based Smart Online Proctoring System for Real-Time Malpractice Detection in Remote Examinations

Abstract

The rapid adoption of online education has introduced serious concerns regarding fairness and academic integrity in remote examinations. This work presents a smart AI-enabled proctoring system that performs continuous monitoring of candidates using video, audio, and behavioral signals. The proposed approach is based on a multi-modal fusion strategy that combines outputs from different detection modules into a unified fraud assessment score. The system identifies suspicious patterns such as impersonation, presence of additional individuals, usage of unauthorized devices, abnormal head movement, and irregular audio activity. Experimental evaluation demonstrates an overall accuracy of 94% along with real-time processing performance of 30 frames per second, making the system suitable for scalable deployment.

Introduction

The text describes an AI-based online exam proctoring system designed to prevent cheating during virtual examinations by using real-time monitoring through webcam, microphone, and browser activity tracking.

Traditional invigilation is ineffective in online exams, leading to increased malpractice. To address this, the proposed system uses Artificial Intelligence, computer vision, and audio processing to monitor students remotely and detect suspicious behavior such as impersonation, multiple faces, mobile phone usage, tab switching, abnormal head movement, and unauthorized audio activity.

The system captures live video and audio data and processes it using multiple AI models: LBPH face recognition for identity verification, YOLOv8 object detection for identifying unauthorized objects, browser monitoring for tab switching detection, and audio analysis for detecting suspicious sounds. These components work together to generate a fraud score, which determines whether an activity is suspicious and should be logged.

The methodology includes student registration with face data, continuous monitoring during exams, periodic snapshot verification, and logging of any irregular behavior. A centralized dashboard allows administrators to review flagged activities after exams.

The system is trained using datasets like COCO, face datasets, and custom webcam recordings, and tested on a low-resource system. It achieves real-time performance with 30 FPS, 94% accuracy, and improved efficiency compared to manual and basic AI systems.

Experimental results show that the proposed system outperforms traditional and single-model approaches in both accuracy and speed. It successfully integrates multiple detection techniques into a unified framework, making online examinations more secure, scalable, and fair.

Conclusion

The growing adoption of online education has significantly increased the need for reliable and intelligent proctoring solutions. As remote examinations become more common, ensuring their credibility and fairness has become a critical requirement for educational institutions. AI-based proctoring systems provide an effective solution by enabling automated and continuous monitoring of candidates without heavy dependence on human invigilators. With the advancement of machine learning and computer vision technologies, it is now feasible to develop highly accurate systems capable of detecting various forms of malpractice in real time. An important aspect of such systems is the ability to record and store evidence of suspicious activities. Maintaining detailed logs not only helps in identifying violations but also plays a crucial role in resolving disputes and ensuring transparency in the evaluation process. This evidence-based approach strengthens trust among students and institutions. Furthermore, accessibility is a key factor in modern education systems. Since a large number of students rely on smartphones rather than personal computers, developing a mobile-compatible proctoring system becomes essential. Future work will focus on improving model accuracy under challenging conditions such as low lighting and background noise.

References

[1] N. Yahya et al., \"AI-Enabled Online Exam Proctoring Systems,\" 2025. [2] V.K. Pant, \"AI-Driven Online Exam Proctoring,\" 2025. [3] Heinrich, \"Online Proctoring Review,\" 2025. [4] T. Singh, \"Automated Proctoring Model,\" 2024. [5] A. Tweissi et al., \"Accuracy of AI-Based Proctoring,\" 2022. [6] S. S. Iyengar and R. R. Brooks, “Online Examination Systems: Challenges and Solutions,” International Journal of Advanced Computer Science, vol. 10, no. 2, pp. 45–52, 2020. [7] A. Ullah and S. Khan, “Secure Web-Based Examination System Using Artificial Intelligence,” International Journal of Computer Applications, vol. 177, no. 29, pp. 12–18, 2020. [8] J. Moten Jr., A. Fitterer, E. Brazier, and J. Leonard, “Investigation of Cyber Cheating in Online Education and Its Prevention Techniques,” Journal of Educational Technology Systems, vol. 48, no. 3, pp. 321–345, 2019. [9] H. M. Alessio, N. Malay, K. Maurer, A. J. Bailer, and B. Rubin, “Examining the Impact of Proctoring on Online Test Performance,” Online Learning Journal, vol. 21. no. 1, pp. 146–161, 2017.

Copyright

Copyright © 2026 Atharva Patil, Jay Parmar, Digambar Ratate, Paritosh Pawar, Prof. Mahendra Patil. 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.