<![CDATA[Integrity and honesty are fundamental aspects of education, including the implementation of the Computer-Based Written Examination (UTBK). Conventional exam supervision is considered less effective in monitoring participants’ behavior due to the limitations in human observation capabilities and consistency. This study develops a real-time cheating-detection system based on camera input by integrating the YOLOv8 algorithm with Farnebäck optical flow. The YOLOv8 algorithm identifies participants’ body poses and activities directly from video footage, while Optical Flow analyzes the direction and motion patterns between frames over time. The system is designed to recognize various suspicious poses such as head-turning, bowing, and cheating-related gestures that indicate potential dishonesty. All detection results are automatically recorded in an SQLite database, complete with timestamps and visual evidence. Experimental results show that the system achieves 94.3% accuracy in detecting suspicious movements. The combination of both methods also helps maintain detection stability when keypoints are not consistently captured in some frames. Additionally, the system is equipped with a graphical user interface (GUI) to facilitate easier monitoring and analysis. These results demonstrate that a pose-and-motion analysis-based approach offers an intelligent and efficient solution for enhancing digital supervision of UTBK examinations.]]>
Copyrights © 2026
