<![CDATA[Conventional Closed-Circuit Television (CCTV) systems generally function only as passive surveillance tools and still require continuous human monitoring, which limits their effectiveness in responding promptly to security threats. This study aims to develop an intelligent CCTV-based security system capable of detecting and recognizing individuals in real time by integrating the YOLO (You Only Look Once) and FaceNet algorithms. YOLO is employed to detect humans and faces from video streams, while FaceNet performs face recognition through embedding extraction and comparison with registered user data. The system is implemented as a web-based application using the Flask framework and integrated with the Telegram Bot API to automatically send notifications when an unknown individual is detected. The prototyping method, consisting of requirement analysis, system development, and iterative evaluation, was adopted in this study. Experimental results show that the FaceNet model achieved the highest accuracy of 97.6% under the 60:40 dataset split scenario, while the YOLO model achieved an average inference time of 9.12 ms per frame. Furthermore, white-box testing confirmed that all system functionalities operated as expected. The proposed system is capable of providing active and real-time security monitoring, enabling faster responses to potential security threats in residential environments.]]>
Copyrights © 2026
