Vision Proctor: An Offline AI-Powered Exam Proctoring System Using Computer Vision

The academic honesty is of primary importance in education. institutions, but the conventional proctoring ap- proaches are resourceintensive and liable to human errors. This paper presents Vision One of the privacy-aware offline exam proctoring systems is Proctor. that makes use of computer vision and deep learning to help. invigilators of human when it comes to examinations. Our method unlike the cloudbased surveillance systems handles all the information. logging violations only are not continuous locally. video streams. The system combines var- ious state of the art systems. such techniques as MediaPipe face detector and head. pose estimation, student identification being a face recognition, and YOLOv8 used to detect banned items like mobile phones. Our prototype is able to monitor 5-6 students at the same time, performing behaviour and attendance marking in an automated manner, real time. analysis, and inclusive reporting violation with suspicion. scoring. The effectiveness of the system in learning abnormal behaviors and ensuring a maintenance at the same time is proved by experimental results. sensitive design philosophy that is conducive to education. The architecture is easy to customize and it is modular. deals with issues of ethics that are common in automated proctoring. systems.