Advancing Solar Panel Inspections: UAV-Based Detection of Cracks Using YOLO Algorithms

Renewable energy sources, particularly solar energy, stand out as vital solutions to global energy and environmental challenges. However, defects such as microscopic or macroscopic cracks in photovoltaic (PV) panels significantly reduce energy production efficiency and system reliability. The rapid, accurate, and autonomous detection of such defects in solar power plants is crucial for reducing efficiency losses and extending the lifespan of PV systems. This study explores the use of advanced object detection algorithms integrated with unmanned aerial vehicles (UAVs) for solar panel inspections. It evaluates the performance comparison of YOLOv5 and YOLOv7 algorithms in detecting cracks under various lighting conditions, complex backgrounds, and resolutions. To enhance model generalization, a comprehensive dataset was created, encompassing various scenarios such as panel sizes, angles, and lighting conditions. The results reveal that YOLOv7 outperformed YOLOv5 with an 8.01% improvement in precision, an 11.05% improvement in recall, a 10.56% improvement in mean average precision (mAP), and a 9.57% improvement in the F1 score. The study highlights the importance of using object detection algorithms to improve the efficiency of solar power plants.