[1]刘 柱,董 琴*,杨国宇,等.基于改进 YOLOv5 的铝型材瑕疵检测算法[J].计算机技术与发展,2023,33(10):183-188.[doi:10. 3969 / j. issn. 1673-629X. 2023. 10. 028]
　LIU Zhu,DONG Qin*,YANG Guo-yu,et al.Aluminum Profile Defect Detection Algorithm Based on Improved YOLOv5[J].,2023,33(10):183-188.[doi:10. 3969 / j. issn. 1673-629X. 2023. 10. 028]

点击复制

基于改进 YOLOv5 的铝型材瑕疵检测算法()

分享到：

《计算机技术与发展》[ISSN:1006-6977/CN:61-1281/TN]

卷:

33

期数:

2023年10期

页码:

183-188

栏目:

人工智能

出版日期:

2023-10-10

文章信息/Info

Title:

Aluminum Profile Defect Detection Algorithm Based on Improved YOLOv5

文章编号:

1673-629X(2023)10--0183-06

作者:

刘 柱; 董 琴* ; 杨国宇; 陈朝峰

盐城工学院 信息工程学院,江苏 盐城 224001

Author(s):

LIU Zhu; DONG Qin* ; YANG Guo-yu; CHEN Chao-feng

School of Mechanical Engineering,Yancheng Institute of Technology,Yancheng 224001,China

关键词:

YOLOv5; 铝型材; 注意力机制; 瑕疵检测; 损失函数; 锚框

Keywords:

YOLOv5; aluminum profiles; attention mechanism; defect detection; loss function; anchor

分类号:

TP391

DOI:

10. 3969 / j. issn. 1673-629X. 2023. 10. 028

摘要:

基于铝型材表面瑕疵类别多样,对实时检测快速精准的需求,提出一种基于改进 YOLOv5 的瑕疵检测算法。通过在原始骨干网络的基础上增加新检测层并使用 K-means++　算法改进锚框的生成方式,提升检测尺度,避免忽视低层语义信息。 对铝型材瑕疵数据集离线增强,丰富样本容量;在 Backbone 网络结构中融入新的卷积结构和 E -CBAM 注意力机制,提高网络的特征提取能力的同时降低冗余计算,提升模型检测性能;采用 EIoU Loss 作为整个网络结构的损失函数来加快收敛效率,解决难易样本不平衡的问题。 实验结果表明,在铝型材瑕疵数据集上将改进后 YOLOv5 检测模型与原始YOLOv5 模型进行比较,平均精度 mAP 提升 2. 9 百分点,召回率 Recall 提升 3. 9 百分点,

速度 FPS 达至 45. 8,将近年来的代表性算法 YOLOv3、YOLOv4、SSD、Faster-rcnn 与改进后的检测算法在铝型材瑕疵数据集上进行性能比较,通过综合对比检测精度、检测速度等重要参数证明改进后的 YOLOv5 检测算法更好地兼顾了检测效率和检测精度。 所提方法满足了铝型材工厂生产现场瑕疵检测要求。

Abstract:

In order to meet the requirement of rapid and accurate real-time detection due to various types of defects on aluminum profiles,a defect detection algorithm based?

on improved YOLOv5 was proposed. By adding a new detection layer on the basis of the originalbackbone network and using K-means+ + algorithm to improve the generation mode of anchor box,the detection scale is increased toavoid ignoring the low-level semantic information. Off-line enhancement of aluminum profile defect data set is conducted to enrich thesample size. New convolution structure and E -CBAM attention mechanism are integrated into Backbone network structure to improvefeature extraction capability,reduce redundant calculation,and improve model detection performance. EIoU Loss is adopted as the lossfunction of the whole network structure to accelerate the convergence efficiency and solve the problem of unbalance of difficult and easysamples. The experimental results show that by comparing the improved YOLOv5 detection model with the original YOLOv5 model inthe aluminum profile defect data set, the average precision mAP increases by 2. 9 percentage points, the recall rate increases by 3. 9　percentage points,and the speed FPS reaches 45. 8. Representative algorithms YOLOv3,YOLOv4,SSD,Faster-rcnn proposed in recentyears were compared with the improved detection algorithm on the aluminum profile defect data set. Through comprehensive comparisonof detection accuracy,detection speed and other important parameters,it is proved that the improved YOLOv5 detection algorithm bettertakes into account detection efficiency and accuracy. The proposed method meets the requirement of defect detection in aluminum profilefactory.

常用功能