Ahmed NoorAlishbaWajeeha ArifM Talha2025-12-262025-12-262025https://escholar.umt.edu.pk/handle/123456789/18054Exam cheating is a prevalent issue that undermines fairness in academia. Conventional approaches, where teachers or supervisors monitor students, are not guaranteed to be effective, particularly in large exam rooms. The purpose of this project is to create an AI- based system that assists in automatically detecting cheating. The system monitors students' posture motion while undertaking exams to detect abnormality. Deep learning models are employed by the system to detect four primary positions: normal, left, right, and back. By processing webcam images in real-time, the system is able to identify abnormal movements and notify examiners for investigation. For greater accuracy, we designed our own student body movement dataset rather than using online available data. We gathered, annotated, and processed the dataset very carefully to make it simulate real exam situations. A CNN-based system and image processing algorithms were employed for training to ensure robust movement detection. The system was tested and validated to monitor the system's accuracy and performance. The system is so designed that it can be operated with real-time exam monitoring software. It offers examiners an easy-to-use dashboard through which they can observe warnings if any student is caught attempting suspicious activities. This saves human effort, reduces errors, and provides a fair exam environment for all the students. In the future, we intend to enhance the system further by incorporating more data, employing eye tracking technology, and experimenting with other deep learning models to make it more accurate. We also intend to integrate it with sophisticated exam monitoring software so that greater institutions may utilize it. This AI-powered solution assists with ensuring fairness in exams, avoiding cheating, and making an exam environment more secureenThesis