Automatic Cheating Detection In Exam Hall

Exams are a widely used evaluation tool in educational institutions to assess students' strengths and weaknesses. However, students tend to cheat during physical exams by exchanging papers, using hidden notes, or fulfilling their parents' expectations, among other things. Due to physical limitations, traditional invigilation methods cannot effectively monitor exams while maintaining their integrity. To address this issue, a study proposes an automated method based on computer vision that uses closed- circuit television (CCTV) cameras to detect suspicious behavior during physical exams. The proposed method employs You Only Look Once (YOLOv3) with residual networks as the backbone architecture to inspect cheating. The results demonstrate that the proposed method is reliable and efficient, achieving 88.03% accuracy in detecting cheating in the classroom environment. Overall, this work shows promising results for invigilating students during exams.