PV module positioning based on improved YOLOv8n

Aiming at the current problems of insufficient positioning and recognition accuracy of photovoltaic module positioning, large calculation complexity, and large parameters, the improved algorithm based on YOLO v8n is proposed for the detection of infrared thermal imaging defects of photovoltaic modules. First, the SPPF module in the original YOLO v8n algorithm is improved, and the SimSPPF module is selected to improve the detection speed of the algorithm. Secondly, the C2f_Faster module is introduced to reduce the parameter amount and computational complexity of the model. Finally, the loss function is optimized, and the CIOU is replaced with the DIOU loss function, so that the network converges faster and more accurate. The improved algorithm can reach $95.5 \%$ of the average detection accuracy of the infrared image detection of photovoltaic modules, which is $0.4 \%$ higher than the original YOLOv8n algorithm $95.9 \%$. The detection speed and accuracy are improved compared with the original YOLO v8n is reduced.