YOLOv8-enhanced defect detection for infrared thermography of PV cells

In photovoltaic (PV) power generation systems, effective defect detection of PV cells is critical for ensuring power generation efficiency and system safety. An improved defect detection algorithm is proposed in this study, where a C2F faster block and an Inception Bottleneck Module (IBM) are integrated into the YOLOv8 network, significantly enhancing the network’s ability to detect defects. Furthermore, a custom-designed infrared thermography experimental system was employed to capture infrared images of both normal and defective PV cells, which were used to train and validate the improved model. An offline data augmentation method was applied to address the limitations of insufficient image datasets. Experimental results demonstrate that the algorithm achieves a mean average precision (mAP) of 97.4%, the detection speed remains at 34 frames per second, effectively balancing detection accuracy and speed. Moreover, comparative analyses with four other models further verify the superior performance of this model, providing valuable insights into the automation and intelligent defect detection of PV cells.