ARCHIVES

The increasing demand for secure and transparent examination systems has highlighted the limitations of conventional human invigilation, particularly in large examination halls. Manual monitoring is prone to fatigue, limited visibility, and delayed response, making it difficult to detect sophisticated cheating behaviors in real time. This paper presents an AI-Based Cheating Detection System for Examination Monitoring that integrates computer vision, deep learning, and edge computing with low-cost IoT hardware to provide intelligent and automated exam surveillance. The proposed system combines ResNet-10 for robust face detection and counting with YOLOv8 Nano for real-time detection of unauthorized mobile devices. To overcome the blind spots of camera-based monitoring, Infrared (IR) and Passive Infrared (PIR) sensors are integrated through a Raspberry Pi Pico H, creating a hardware–software closed-loop architecture capable of detecting under-desk activities and abnormal movements. Live video streams captured through a webcam and ESP32-CAM are processed using OpenCV and displayed on a Flask-based monitoring dashboard. The system evaluates six predefined cheating scenarios, including face absence, extra person detection, student proximity, unauthorized device usage, abnormal movement, and under-desk activity, while simultaneously triggering localized buzzer alerts for immediate intervention. Experimental evaluation demonstrates reliable real-time monitoring with visual processing speeds of approximately 15–20 FPS and hardware response latency below 50 ms, ensuring rapid and accurate detection. The proposed architecture is scalable, cost-effective, and suitable for deployment in educational institutions, providing an efficient solution for enhancing examination integrity while reducing the workload of human invigilators.