基于改进YOLOv5s的玻璃盖板划伤检测算法
Glass Cover Screen Scratch Detection Algorithm Based on Improved YOLOv5s
李虎 1胡晓兵 1陈海军 1毛业兵 1章程军 1李航1
作者信息
- 1. 四川大学机械工程学院,成都 610065;四川大学宜宾产业技术研究院,宜宾 644000
- 折叠
摘要
针对当前玻璃盖板检测速度较慢、精确率较低的问题,提出一种基于YOLOv5s算法的玻璃盖板划伤检测改进模型.首先,借鉴ResNeXt结构和大核注意力(large kernel attention,LKA)结构改进原C3 模块,增强网络对于特征的检测和提取能力;其次,向网络中引入BiFPN模块,提高网络的特征融合能力和小目标检测能力;最后,使用EIOU损失函数替换原网络中的CIOU损失函数,提高锚框生成的准确性和模型收敛速度.结果表明,改进后模型,精确率达到 98.2%,召回率达到98.4%,实现玻璃盖板划伤的高效检测.
Abstract
In order to realize the efficient automatic detection of glass cover screen,aiming at the problem of slow speed and low accuracy of manual detection,an improved model of scratch detection of glass cover screen based on YOLOv5s algorithm is proposed.Firstly,the original C3 module is improved by referring to the ResNeXt structure and the large kernel attention(LKA)structure to enhance the network's ability to detect and extract features.Secondly,the BiFPN module is introduced into the network to improve the fea-ture fusion ability and small target detection ability.Finally,the EIOU loss function is used to replace the CIOU loss function to improve the accuracy of anchor generation and the convergence speed.The results show that the accuracy rate of the improved model reaches 98.2%,and the recall rate reaches 98.4%,which realizes the high-precision detection of glass cover screen scratches.
关键词
划伤检测/YOLOv5s/ResNeXt/大核注意力/智能制造Key words
scratch detection/YOLOv5s/ResNeXt/large kerneal attention/intelligent manufacturing引用本文复制引用
基金项目
川大-宜宾校市合作项目(2020CDYB-3)
四川省科技计划重点研发项目(2022YFG0072)
四川大学-遂宁校市合作专项资金项目(2020CDSN-04)
出版年
2024
NETL
NSTL
维普
万方数据