A new small target defect detection algorithm for solar panels based on improved YOLOV7

Surface Defect Detection (SDD) is an important means to ensure product quality in industrial production. Accompanied by the rapid development of artificial intelligence technology, the surface defect detection algorithm based on convolutional neural network (CNN) has gradually replaced the traditional machine vision and manual inspection methods. However, due to the small morphology of some defects in solar panels, this brings certain challenges to industrial defect detection. Therefore, this paper investigates a surface defect detection based on YOLOV7 for the small target in solar panels. Firstly, the coordinate attention mechanism is used to improve the model’s attention to distinguishable defect features. Secondly, a two-way weighted feature pyramid is utilised to replace the PANet in order to improve the feature fusion effect of the detection model. Finally, Soft Non-Maximum Suppression (Soft NMS) is used in post-processing to reduce the leakage rate of the detection model. The proposed method is tested on a self-built dataset of surface defects detection on solar panels, and the experimental results verify that the accuracy of the model in this paper is better than other advanced methods.