A Model for Detecting Suspicious Behaviors During Examinations Using Deep Learning focuses on building a smart system that can automatically monitor students and detect cheating-related actions during exams. Examination malpractice remains a serious challenge in educational institutions, as human invigilators may fail. This reduces the credibility in the examination process. To address this problem, the project will developed a deep learning model that uses Convolutional Neural Networks (CNN) to capture visual features from video frames and Long Short-Term Memory (LSTM) networks to analyze behavior patterns over time. The solution will be guided by two main methodologies: CRISP-DM, which structured the data mining process through phases; and OOADM, which will guide the system design using object-oriented principles. The system will be implemented using Python as the programming language and Django for database development and management. This model is expected to effectively identify suspicious behaviors with high accuracy.

Keywords: Frame, Image, Machine Learning, Digital Image and Deep Learnig