Suspicious Activity Detection in Exam Hall using Deep Learning

Abstract: Cheating in academic examinations compromises the integrity and fairness of educational assessments. Manual invigilation alone is often insufficient, especially in large-scale exams, where continuous monitoring is impractical. This study proposes a real-time, deep learning-based system for detecting suspicious activity in examination halls using surveillance footage. By leveraging Convolutional Neural Networks (CNNs) and Long Short-Term Memory (LSTM) networks, the system analyzes student behavior across video frames to flag anomalies such as unusual head movements, note passing, or collaboration. The solution enhances accuracy through temporal behavior analysis and operates non-intrusively using standard CCTV setups. Designed for scalability and minimal human oversight, the system aims to elevate examination security while maintaining student privacy. Keywords: Exam Surveillance, Deep Learning, CNN-LSTM, Suspicious Activity Detection, Computer Vision, Real- Time Monitoring, Academic Integrity.