现代电子技术2025，Vol.48Issue(11)：144-150,7.DOI:10.16652/j.issn.1004-373x.2025.11.022
基于改进YOLOv8的印刷电路板缺陷检测模型
Printed circuit board defect detection model based on improved YOLOv8
摘要
Abstract
In view of the current PCB(printed circuit board)defect detection problems such as excessively small objects,low detection accuracy and slow detection speed,a PCB defect detection model based on improved YOLOv8 is proposed.In the model,YOLOv8n is taken as the framework.The RFCBAMConv(receptive-field concentration-based attention module convolutional operation)module,which is generated by combining the RFAConv(receptive-field attention convolutional operation)and CBAM(concentration-based attention module)attention mechanism,is introduced to improve the feature extraction ability of the backbone network.The TFE(triple feature encoding)and SSFF(scale sequence feature fusion)module in the small object segmentation algorithm ASF-YOLO are introduced to improve the neck network.A small object detection head that fuses multi-scale features is constructed in combination with the SSFF module.The NWD(normalized Wasserstein distance)loss function is used to optimize the defects of small object recognition.The experimental results show that the accuracy rate,recall rate,mAP@0.5 and mAP@0.5:0.95 of the improved model is enhanced by 1.7%,2.9%,1.8%and 3.8%,respectively,in comparison with those of the original model,its model size is reduced by 10%,and its number of parameters is reduced by 13%.To sum up,the proposed model can be applied to the PCB defect detection task effectively.关键词
缺陷检测/PCB/感受野注意力/损失函数/YOLOv8/小目标Key words
defect detection/PCB/receptive-field attention/loss function/YOLOv8/small object分类
电子信息工程引用本文复制引用
范博淦,王淑青,陈开元..基于改进YOLOv8的印刷电路板缺陷检测模型[J].现代电子技术,2025,48(11):144-150,7.基金项目
国家自然科学基金项目:构件复杂背景下的实景三维古建筑物细节多层次语义提取方法研究(62306107) （62306107）
