Computer Vision–Based Malpractice Detection System

Examinations are the most important aspects in student lives. In order to secure high score, they may adopt different cheating approaches during the examinations. Offline examinations have human invigilators to monitor students, who have their own physical limitations. This study aims to solve the exam’s malpractice related issues through computer vision. Computer vision is a subfield of artificial intelligence and machine learning, computer vision becomes one of the hottest fields with its extensive variety of applications and to imitate the commanding capabilities of human vision. We develop a model using computer vision to detect physical behavior like facial expressions, neck movements, body poses and suspicious activities like exchanging sheets, hiding notes. We will develop a model which can also detect prohibited gadgets during the examination. The proposed method uses You Only Look Once (YOLO) algorithm with residual networks as the back bone architecture to inspect cheating in exams through cameras.