Abstract—Most Educational institutions still rely on Traditional Closed-Circuit Television (CCTV) systems with manual monitoring, which limits their effectiveness in real-time surveillance. Furthermore, deploying advanced surveillance technologies in environments such as universities, colleges, and schools is challenging due to high computational requirements, cost inefficiency, and scalability constraints. To address these limitations, this study proposes a lightweight surveillance framework that uses FaceNet for face recognition, Multi-task Cascaded Convolutional Neural Networks (MTCNN) for face localization during training, and Haar Cascade for real-time face localization during deployment. The system processes CCTV footage by converting videos into frames, detecting faces using the Haar Cascade algorithm, and generating facial embeddings with FaceNet. These embeddings are then classified using Machine Learning (ML) and Deep Learning classifiers (DL) to identify individuals as either authorized or unauthorized. Evaluations conducted on a custom dataset of 20 students around 2000 images created explicitly for university scenarios demonstrated a recognition accuracy of 97.1% using Support Vector Machine (SVM), with the lowest inference time, and included a user-friendly Graphical User Interface (GUI) developed with Flask for real-time monitoring. This approach offers a scalable, computationally efficient, cost-effective, and practical solution for enhancing campus security by effectively identifying unauthorized individuals.
 
Keywords—deep learning, Closed-Circuit Television (CCTV) surveillance, education sector security, university campus security, FaceNet, lightweight security system


Copyright © 2025 by the authors. This is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited (CC BY 4.0).