Cheating detector - Real-Time AI-Driven Exam Proctoring Engine

Abstract: With the rapid expansion of online education and remote assessments, maintaining academic integrity without human intervention has become a critical challenge. This paper presents Cheating_detector, a real-time, AI-driven automated exam proctoring engine. The system integrates computer vision techniques utilizing Python, OpenCV, and YOLOv8 to perform concurrent object detection, pose estimation, and gaze tracking. By analyzing candidate behavior in real time, the engine identifies suspicious activities such as forbidden object usage, improper posture, and unauthorized visual deviations. Furthermore, an automated alert and logging mechanism built with SQLite3 and an SSL-encrypted smtplib module ensures reliable incident storage and dispatches rate-limited, evidence-backed email notifications to administrators. Experimental evaluation demonstrates high real-time processing performance, minimal false-positive rates, and robust incident logging for academic integrity enforcement.

Keywords: Automated Exam Proctoring, Computer Vision, YOLOv8, Real-Time Object Detection, Pose Tracking, Academic Integrity.