Post by MDPI Engineering

📢 #CallForReading | #Article in Automation MDPI 📹 AI for Academic Integrity: GPU-Free Pose Estimation Framework for Automated Invigilation 🎓 By Syed Muhammad Sajjad Haider, Muhammad Zubair, Aashir Waleed, Muhammad Shahid, Furqan Asghar, Muhammad Omer Khan 🔗 Full Paper: https://lnkd.in/gQKuQEvC #ComputerVision #YOLOv8 #CheatingDetection #AI Educational exams rely on credible evaluation, yet manual invigilation fails to fully curb academic cheating. Leveraging AI and computer vision, this study develops a lightweight, real-time exam cheating detection system based on YOLOv8 object detection and human pose estimation. It innovatively constructs two geometric triangles via facial key point distances to calculate facial pose angles, adopting dynamic frame thresholds to identify suspicious cheating behaviors. All violation records are automatically logged and stored. Requiring no GPU support and featuring low computational cost, the system outperforms previous methods, achieving 96.18% accuracy and 96.2% precision in actual exam scenarios.