Single Object Tracking for Defense Applications

Single object tracking (SOT) is essential in defense applications, facilitating the ongoing surveillance of objects like vehicles, drones, or humans in dynamic settings. This work introduces a methodology for SOT utilizing the YOLOv11 object detection model, tailored for real-time efficacy and precision. Utilizing the VisDrone SOT dataset, which comprises varied and intricate aerial imagery, our approach proficiently monitors objects under demanding circumstances, including occlusion, fluctuating illumination, and swift movement. YOLOvl1's design, distinguished by its rapidity and accuracy, is optimized to improve tracking precision while ensuring minimal latency, crucial for practical defense applications. The experimental results illustrate the model's robustness in terms of accuracy and computing efficiency when applied to VisDrone SOT data, rendering it appropriate for surveillance, reconnaissance, and autonomous systems. The proposed approach underscores the efficacy of YOLOv11 as a dependable instrument for real-time defense-oriented object tracking activities.