• 中国计算机学会会刊
  • 中国科技核心期刊
  • 中文核心期刊

计算机工程与科学 ›› 2023, Vol. 45 ›› Issue (07): 1226-1235.

• 图形与图像 • 上一篇    下一篇

基于改进YOLOv3的金属表面缺陷检测

刘浩翰,孙铖,贺怀清,惠康华   

  1. (中国民航大学计算机科学与技术学院,天津 300300)
  • 收稿日期:2021-10-18 修回日期:2022-01-08 接受日期:2023-07-25 出版日期:2023-07-25 发布日期:2023-07-11
  • 基金资助:
    国家重点研发计划(2020YFB1600101);天津市教委科研项目(2020KJ024);

Metal surface defect detection based on improved YOLOv3

LIU Hao-han,SUN Cheng,HE Huai-qing,HUI Kang-hua   

  1. (College of Computer Science and Technology,Civil Aviation University of China,Tianjin 300300,China)
  • Received:2021-10-18 Revised:2022-01-08 Accepted:2023-07-25 Online:2023-07-25 Published:2023-07-11

摘要: 为提高工业零件表面缺陷的检测效率,提出一种基于改进YOLOv3的目标检测方法。引入目前最新的具有通道置换的注意力机制SA,将其与YOLOv3模型骨架结构Darknet-53的残差单元进行组合改进构成SA残差块结构,充分利用特征通道信息,得到YOLOv3-SA模型。针对不同数据集,对输入图像进行不同规模比例缩放,分别使用K-means方法对真实标框进行聚类以提高检测效率。实验结果表明,YOLOv3-SA模型查全率达95.4%,相比YOLOv3,mAP最多可提高约7%。

关键词: 深度学习, 目标检测, YOLO, Shuffle, SA, 注意力机制, K-means

Abstract: In order to improve the efficiency of detecting surface defects on industrial parts, a target detection method based on improved YOLOv3 is proposed. The latest attention mechanism SA (Shuffle Attention) with channel shuffling is introduced and combined with the residual unit of the Darknet-53 backbone structure of the YOLOv3 model to form the SA residual block structure, which fully utilizes the feature channel information to obtain the YOLOv3-SA model. For different datasets, the input images are scaled at different scales, and the K-means method is used to cluster the real bounding boxes to improve detection efficiency. The experimental results show that the recall rate of the YOLOv3-SA model reaches 95.4%, and the mAP can be increased by up to 7% compared to YOLOv3.

Key words: deep learning, target detection, YOLO, Shuffle;shuffle attention, attention mechanism, K-means