基于改进YOLOv8的PCB焊点语义分割方法OA

Aiming at the problem of false detection and missing detection of Printed Circuit Board(PCB)solder joints with similar gray values in the process of detection and segmentation,a semantic segmentation model for PCB solder based on improved YOLOv8 is proposed.Firstly,Coordinate Attention(CA)mechanism is introduced into the backbone network to accurately locate the spatial position of solder joints and to improve the ability of the model to capture the spatial information of solder joints.Secondly,Bi-directional Feature Pyramid Network(BiFPN)is used to replace the Path Aggregation Network(PANet)feature pyramid to better capture the boundary information of the target,and add a segmentation layer on the basis of the original.Finally,the EIoU loss function is introduced to provide finer evaluation results and improve generalization.Comparative experiments show that the mean Pixel Accuracy(mPA),mean Intersection over Union(mIoU)and Frames Per Second(FPS)of the proposed algorithm reach 90.37％,83.76％and 43 FPS,which realizes more accurate solder joint segmentation of PCB board.