<![CDATA[This study presents a real-time human detection and face recognition system that utilizes CCTV video streams and a localhost-based monitoring dashboard. Using a Research and Development (R&D) approach, a computer vision application was designed and implemented on a laptop platform. Human detection was performed using YOLO11n, facial regions were localized with YuNet, and face recognition was carried out using SFace to distinguish registered individuals from unknown persons. The detection results were displayed through a web-based dashboard that provided live video streaming, AI status information, face registration, and detection history records. Performance evaluation was conducted under various conditions, including human and non-human scenarios, registered and unknown faces, dark-room environments, and night-vision mode. The dashboard maintained a preview rate of approximately 20–30 FPS. Experimental results showed that human detection achieved an accuracy of 80%, while face recognition achieved 72% accuracy under the tested conditions. Alert Level 1 was triggered when a person was detected, whereas Alert Level 2 was activated for unknown-face events. The findings demonstrate the potential of integrating lightweight computer vision models into a local surveillance system without relying on cloud infrastructure. Nevertheless, system performance remained dependent on factors such as lighting conditions, camera distance, face orientation, image quality, and available computing resources.]]>
