铝制型材表面缺陷检测算法的研究  

Study of Screen Print Pattern Defect Detection Algorithm

作　　者：李星炎 张勇斌[1] 付秀丽[2] LI Xingyan;ZHANG Yongbin;FU Xiuli(Beijing Institute of Graphic Communication,Beijing 102600,China;Beijing Institute of Petrochemical Technology,Beijing 102617,China)

机构地区：[1]北京印刷学院,北京102600 [2]北京石油化工学院,北京102617

出　　处：《北京印刷学院学报》2023年第9期14-20,共7页Journal of Beijing Institute of Graphic Communication

摘　　要：目前针对铝制型材的表面缺陷检测手段主要是人工检测,存在主观性强、检测效率低等问题。为了提高铝制型材的表面缺陷检测效率以及准确率,提出了一种新的检测算法Yolov5-GCE,该算法融合Yolov5算法、CBAM卷积注意力模块和GhosttNet模块,使用了改进后的EIoU损失函数,提高了算法精度。以天池竞赛的铝制型材表面缺陷数据集为验证数据通过与改进前算法做对比,结果显示该算法的平均精度提高了3.8%,表明本文算法具有更好的检测能力。At present,the surface defect detection means for aluminum profiles is mainly manual detection,which has the problems of strong subjectivity and low detection efficiency.In order to improve the efficiency and accuracy of surface defect detection for aluminum profiles,a new detection algorithm,Yolov5-GCE,is proposed,which combines the Yolov5 algorithm,the CBAM convolutional attention module and the GhosttNet module,and uses the improved EIoU loss function to improve the algorithm accuracy.Taking the aluminum profile surface defects dataset from the Tianchi competition as the validation data by making a comparison with the pre-improvement algorithm,it shows that the average accuracy of the algorithm has been improved by 3.8%,and the results show the better detection ability of the algorithm in this paper.

关 键 词：缺陷检测 深度学习 Yolov5-GCE 

分 类 号：TP39[自动化与计算机技术—计算机应用技术]