DRONE-ASSISTED REAL-TIME PERSON DETECTION AND ZONE MONITORING SYSTEM

This paper presents a real-time person detection and restricted zone monitoring system for deployment on an operator-controlled quadcopter drone. YOLOv8 is used for per-frame person detection; DeepSORT maintains persistent individual identities across frames. Restricted zones are defined as pixel-coordinate polygons and evaluated each frame using point-in-polygon analysis. Confirmed zone breaches trigger audio-visual alerts and, if enabled, an autonomous follow mode. A modular architecture decouples the video source from the processing pipeline, enabling transition from a smartphone to the DJI Tello drone without changes to detection or monitoring logic. Initial validation confirms reliable person detection across varying distances and environments.