摘要
发电厂厂区内违规吸烟易导致火灾、爆炸等事故,会带来巨大损失;针对电厂内人员违规吸烟行为检测精度不高的问题,提出一种基于改进YOLOv5s(You Only Look Once v5s)的电厂内人员违规吸烟检测方法;该方法以YOLOv5s网络为基础,将YOLOv5s网络C3模块Bottleneck中的3×3卷积替换为多头自注意力层以提高算法的学习能力;接着在网络中添加ECA(Efficient Channel Attention)注意力模块,让网络更加关注待检测目标;同时将YOLOv5s网络的损失函数替换为SIoU(Scylla Intersection over Union),进一步提高算法的检测精度;最后采用加权双向特征金字塔网络(BiFPN,Bidirectional Feature Pyramid Network)代替原先YOLOv5s的特征金字塔网络,快速进行多尺度特征融合;实验结果表明,改进后算法吸烟行为的检测精度为89.3%,与改进前算法相比平均精度均值(mAP,mean Average Precision)提高了2.2%,检测效果显著提升,具有较高应用价值。
Illegal smoking easily leads to the accidents of fire and explosion in power plants,it will bring huge loss to the people.Aiming at the problem that the detection accuracy of illegal smoking behavior of personnel is not high in the power plants,an illegal smoking detection method of personnel is proposed on the basis of improved YOLOv5s in the power plants.The method is based on YOLOv5s network,the 3×3 convolution in the C3 module of the YOLOv5s network with a Bottleneck layer is replaced to improve the learning ability of the algorithm.Then the efficient channel attention(ECA)module is added to the network,which makes the network pay more attention to the target to be detected.At the same time,the loss function of the YOLOv5s network is replaced by Scylla intersection over union(SIoU)to further improve the detection accuracy of the algorithm.Finally,the weighted bidirectional feature pyramid network(BiFPN)is used to replace original YOLOv5s feature pyramid network to rapidly perform multi-scale feature fusion.The experimental results show that,compared with the original algorithm,the detection accuracy of the improved algorithm is 89.3%,and the mean average precision(mAP)is increased by 2.2%.Its detection effect improves significantly,and it has high application value.
作者
王彦生
曹雪虹
焦良葆
孙宏伟
高阳
WANG Yansheng;CAO Xuehong;JIAO Liangbao;SUN Hongwei;GAO Yang(AI Industrial Technology Research Institute,Nanjing Institute of Technology,Nanjing 211167,China;Jiangsu intelligent perception technology and equipment Engineering Research Center,Nanjing 211167,China)
出处
《计算机测量与控制》
2023年第5期48-55,共8页
Computer Measurement &Control
基金
江苏省自然科学基金项目(BK20201042)
江苏省政策引导类计划项目(SZ2020007)。
关键词
吸烟
目标检测
多头自注意力层
注意力模块
损失函数
加权双向特征金字塔
smoking
target detection
multi-self attention layer
attention module
loss function
weighted bidirectional feature pyramid
