DYNAMIC PRESENCE TRACKING SYSTEM IN SHARED ENVIRONMENTS

This research presents an AI-driven Dynamic Presence Tracking System designed to enhance security, access control, and attendance management in restricted environments. Traditional surveillance systems suffer from high false-positive rates, latency, and an inability to distinguish between authorized and unauthorized individuals, making them inefficient for real-time monitoring. To address these challenges, our proposed system integrates YOLOv11 for real-time object detection, FaceNet for facial recognition, and DeepSORT for persistent tracking across camera networks. By leveraging edge computing on NVIDIA Jetson Xavier, the system processes live video feeds at 30 FPS, achieving 96.2% tracking accuracy on the MOT17 benchmark. A customized alarm mechanism provides context-aware alerts for unauthorized access, while automated CSV logs track entry and exit times for audit purposes. Compared to traditional motion-sensor-based systems, our approach reduces false positives by 42% and effectively mitigates issues related to occlusion and varying lighting conditions. The system’s scalability and adaptability make it a robust framework for real-time surveillance, security enhancement, and operational efficiency in both corporate and public environments. Through optimized deep learning models and edge processing, our work bridges critical gaps in modern surveillance technologies, offering a more intelligent and reliable presence-tracking solution.