用户登录
期刊信息
  • 主管单位:
  • 上海市教育委员会
  • 主办单位:
  • 上海理工大学
  • 主  编:
  • 庄松林
  • 地  址:
  • 上海市军工路516号
  • 邮政编码:
  • 200093
  • 联系电话:
  • 021-55277251
  • 电子邮件:
  • xbzrb@usst.edu.cn
  • 国际标准刊号:
  • 1007-6735
  • 国内统一刊号:
  • 31-1739/T
  • 邮发代号:
  • 4-401
  • 单  价:
  • 15.00
  • 定  价:
  • 90.00
朱文博,夏林聪,陈龙,吴晨睿,陈红光.基于改进YOLOv5的O型密封圈缺陷检测方法[J].上海理工大学学报,2022,44(5):440-448.
基于改进YOLOv5的O型密封圈缺陷检测方法
Defect detection method of O-ring based on improved YOLOv5
投稿时间:2022-04-13  
DOI:10.13255/j.cnki.jusst.20220413001
中文关键词:  YOLOv5  O型密封圈  缺陷检测  卷积注意力机制  双向特征金字塔网络
英文关键词:YOLOv5  O-ring  defect detection  convolutional block attention  bidirectional feature pyramid network
基金项目:国家自然科学基金资助项目(52075340,52105525)
作者单位E-mail
朱文博 上海理工大学 机械工程学院上海 200093  
夏林聪 上海理工大学 机械工程学院上海 200093  
陈龙 上海理工大学 机械工程学院上海 200093 cl@usst.edu.cn 
吴晨睿 上海理工大学 机械工程学院上海 200093  
陈红光 上海贝特威自动化科技有限公司上海 201109  
摘要点击次数: 0
全文下载次数: 1
中文摘要:
      针对O型密封圈缺陷难以人工识别的问题,提出一种基于改进YOLOv5的表面缺陷自动检测方法。在数据预处理阶段,采用半自动标注方法减少人工标注成本,同时将拼接图片改为9张以实现Mosaic数据增强方法。在网络预测层引入标签平滑方法以减少模型过度依赖标签。在骨干网络中添加卷积注意力机制模块,强化有效信息,使骨干网络提取更加细致的局部特征信息。同时,针对缺陷类型尺度变化大的特点,引入剪枝的双向特征金字塔网络,以解决大小缺陷在特征提取过程中的丢失问题。实验结果表明,基于改进的YOLOv5与原YOLOv5相比,O型圈表面缺陷检测平均精度均值提高了4.26%,并且检测速度在25 ms之内,能够满足实际生产需要。
英文摘要:
      Aiming at the difficulty of manual identification of O-ring defects, an automatic detection method of surface defects based on the improved YOLOv5 was proposed. In the data preprocessing stage, the semi-automatic labeling method was used to reduce the cost of manual labeling, and the number of mosaic images used was changed to nine to realize the mosaic data enhancement method. A label smoothing method was introduced in the network prediction layer to reduce the model's over-reliance on labels. The convolutional attention mechanism module was added to the backbone network to highlight the valid information, so that the backbone network could extract more detailed local feature information. At the same time, because of the characteristics of large scale changes of defect types, a pruned bidirectional feature pyramid network was introduced to tackle the loss of both large and small defects in the feature extraction process. The experimental results show that compared with the original YOLOv5, the average accuracy of O-ring surface defect detection in the improved YOLOv5 is increased by 4.26%, and the detection speed is within 25 ms, which can meet the actual production needs.
HTML   查看全文  查看/发表评论  下载PDF阅读器