Automated Invigilation System for Offline Exam Monitoring Using Deep Learning

Exam proctoring is an exhausting job since it gets more diligently for the supervisors to keep an eye on the students in the testing area. In the field of education, determining suspicious conduct during exams is crucial because it tackles the vital problem of upholding academic integrity. It benefits both educational institutions and students by fostering an environment that is fair and deserving of trust for assessments. The majority of the current methods depends on traditional invigilation systems, which are difficult to use and mainly rely on human supervisors. They not only require a large amount of labor, which raises costs, but also a great deal of coordination and training. In this work, a more dependable, effective, and technologically advanced method for tracking and evaluating activities during offline examination is presented. The proposed system uses the internet of things (lOT) combined with sophisticated deep learning techniques to identify and detect suspicious activity and noisy hot-spots. For training and evaluation, we utilized benchmark dataset, xamDataSetv2, for our experiments and achieved 98%, 73%, 91%, and 80.6% of accuracy, precision, recall, and fl-score respectively.