Autonomous UAV-Based Monitoring of Fuel Tanks Using YOLOv8 for Global Energy Management

In this study, we investigate the detection of fuel tanks, critical components of global energy management and infrastructure security, from UAV-acquired images using the YOLOv8 algorithm. The synergy between UAVs and the YOLOv8 object detector is highlighted, demonstrating its capacity to automate detection and monitoring processes for industrial control applications. YOLOv8's advanced architecture, featuring a robust backbone and efficient neck design, enables accurate detection of objects across various sizes and conditions, even within complex scenes. As the latest iteration of the YOLO series, YOLOv8 surpasses its predecessors in accuracy and processing speed, incorporating advanced features that enhance its detection capabilities. The study evaluates YOLOv8's performance in detecting fuel tanks under diverse scenarios. Key results include a sensitivity rate of 0.888, indicating high precision in positive predictions, and a recall rate of 0.896, reflecting a low target miss rate. The mean average precision (mAP) of 0.891 and F1 score of 0.892 underscore the algorithm's balanced optimization of accuracy and sensitivity. Additionally, YOLOv8 achieves a rapid processing time of 41 ms per image, highlighting its suitability for real-time applications. These findings contribute significantly to the adoption of innovative technologies in energy and industrial control sectors, demonstrating YOLOv8's effectiveness and reliability for automated monitoring tasks.