Detecting and Monitoring UAVs Using YOLOv 8 and Tracking Algorithms

The rapid proliferation of unmanned aerial vehicles (UAVs) in applications such as environmental monitoring, agriculture, and cinematography has heightened concerns regarding safety and privacy. To address these challenges, this paper proposes a real-time UAV detection and tracking system leveraging the YOLOv8 model, optimized for high-accuracy recognition of small and fast-moving UAVs in aerial imagery. YOLOv8, an advanced object detection framework, is enhanced with dynamic anchor selection, an improved backbone network, and optimized loss functions to achieve superior performance in UAV detection. The system integrates the Simple Online and Realtime Tracking (SORT) algorithm, which combines the Kalman filter for trajectory prediction and the Hungarian algorithm for data association, ensuring robust and continuous tracking of UAVs across video frames. Experimental results on a custom UAV dataset demonstrate the system's effectiveness, with high precision, recall, and mean Average Precision (mAP) metrics. The integration of YOLOv8 and SORT significantly reduces ID switching and detection fragmentation, enabling reliable realtime UAV monitoring in dynamic environments. This work highlights the potential of the proposed system for security and surveillance applications, with future research directions including occlusion handling, multi-UAV tracking, and deployment on edge devices.