|本期目录/Table of Contents|

[1]魏振强,马永刚,曹欣宜.基于改进YOLOv5的石化设备目标检测算法技术研究*[J].中国安全生产科学技术,2025,21(1):139-145.[doi:10.11731/j.issn.1673-193x.2025.01.018]
　WEI Zhenqiang,MA Yonggang,CAO Xinyi.Research on target detection algorithm technology of petrochemical equipment based on improved YOLOv5[J].JOURNAL OF SAFETY SCIENCE AND TECHNOLOGY,2025,21(1):139-145.[doi:10.11731/j.issn.1673-193x.2025.01.018]
点击复制

基于改进YOLOv5的石化设备目标检测算法技术研究*
分享到：

《中国安全生产科学技术》[ISSN:1673-193X/CN:11-5335/TB]

卷:
21
期数:
2025年1期
页码:
139-145
栏目:
职业安全卫生管理与技术
出版日期:
2025-01-30

文章信息/Info

Title:
Research on target detection algorithm technology of petrochemical equipment based on improved YOLOv5
文章编号:
1673-193X(2025)-01-0139-07
作者:
魏振强; 马永刚; 曹欣宜
（1.中国石油集团安全环保技术研究院有限公司，北京 102206;
2.中国石油四川石化有限责任公司，四川 成都 611939）
Author(s):
WEI Zhenqiang;  MA Yonggang;  CAO Xinyi
(1.CNPC Research Institute of Safety & Environment Technology,Beijing 102206,China;
2.CNPC Sichuan Petrochemical Co.,Ltd.,Chengdu Sichuan 611939,China)
关键词:
YOLOv5; 注意力机制; 石化设备; 设备数据集; 目标检测
Keywords:
YOLOv5;  attention mechanism;  petrochemical equipment;  equipment dataset;  target detection
分类号:
X937
DOI:
10.11731/j.issn.1673-193x.2025.01.018
文献标志码:
A
摘要:
为解决石化设备目标检测任务中，由于设备种类多、尺度跨度大和背景复杂等原因，造成设备目标检测误检率和漏检率较高的问题，研究多个CCbam3层、通道和空间注意机制、SiLU激活函数等对YOLOv5模型的作用机制，提出1种单阶段注意力机制增强的改进YOLOv5模型，该模型主干模块改进为焦点层、卷积层、多个CCbam3层和空间金字塔池化层；中间模块则使用通道和空间注意机制模块来优化不同设备尺度的特征图；此外，该模型使用SiLU作为激活函数，提高网络的自稳定性；最后，预测模块采用4个检测模块来检测不同尺寸的设备。研究结果表明：改进的YOLOv5模型在真实的设备数据集获得的P-R曲线平均精度为97.8%，石化设备均得到精确地检测和定位；在mAP50方面，较YOLOv5s，YOLOv5x分别提升3.2和0.4百分点；添加的注意力机制的模型有利于提取石化设备特征并更准确地进行目标检测。研究结果可为石化设备的智能化目标检测和过程安全风险防控工作提供技术支撑。
Abstract:
In the target detection of petrochemical equipment,due to the variety of equipment types,large size spans,and complex backgrounds,the false detection rate and missed detection rate of equipment target detection are relatively high.To address the aforementioned issues,the action mechanism of multiple CCBam3 layers,channel and spatial attention mechanism,and SiLU activation function on the YOLOv5 model was investigated,and an improved YOLOv5 model with single-stage attention mechanism enhancement was proposed.The backbone module of the model were improved to the focus layer,the convolutional layer,multiple CCbam3 layers,and the spatial pyramid pooling layer.The intermediate module employed the channel and spatial attention mechanism module to optimize the feature maps for different equipment scales.In addition，the model used SiLU as the activation function to improve the self-stability of the network.Finally,the prediction module adopted four detection modules to detect the equipment with different sizes.The results show that the improved YOLOv5 model achieves an average P-R curve accuracy of 97.8% on the real equipment datasets,and all the petrochemical equipment are accurately detected and located.In terms of mAP50,it improves by 3.2 and 0.4 precentage points compared to YOLOv5s and YOLOv5x.The added attention mechanism model is beneficial for extracting the features of petrochemical equipment and detecting the targets more accurately.The research results can provide technical support for the intelligent target detection and process safety risk prevention and control of petrochemical equipment.

参考文献/References:

［1］HUANG Z,YANG S,ZHOU M,et al.Making accurate object detection at the edge:review and new approach［J］.Artifcial Intelligence Review,2022,55(3):2245-2274.
［2］KANG J,TARIQ S,OH H.A survey of deep learning-based object detection methods and datasets for overhead imagery［J］.IEEE Access,2022,10:20118-20134.
［3］REDMON J,DIVVALA S,GIRSHICK R,et al.You only look once:unifed,real-time object detection［C］//The IEEE Conference on Computer Vision and Pattern Recognition,2016.
［4］REDMON J，FARHADI A.YOLO9000:better,faster,stronger［C］//The IEEE Conference on Computer Vision and Pattern Recognition,2017.
［5］邵延华,张铎,楚红雨，等.基于深度学习的YOLO目标检测综述［J］.电子与信息学报,2022,44(10):3697-3708. SHAO Yanhua，ZHANG Duo,CHU Hongyu,et al.A review of YOLO object detection based on deep learning［J］.Journal of Electronics & Information Technology,2022,44(10):3697-3708.
［6］BOCKKOVSKIY A,WANG C Y,LIAO H Y M.YOLOv4:optimal speed and accuracy of object detection［EB/OL］.(2020-04-23)［2025-01-06］.https://arxiv.org/abs/2004.10934.
［7］JOCHER G,STOKEN A,BOROVEC J.Ultralytics/yolov5:V4.0-Nn.SiLU［M］.2021.
［8］LIU W,ANGUELOV D,ERHHAN D,et al.SSD:single shot multibox detector［C］//Computer Vision-ECCV 2016,2016.
［9］FU C Y,LIU W,RANGA A,et al.DSSD:Deconvolutional Single Shot Detector［EB/OL］.(2017-01-23)［2025-01-06］.https://arxiv.org/abs/1701.06659.
［10］DUAN K,BAI S,XIE L,and et al.CenterNet:keypoint triplets for object detection［C］//The 2019 IEEE International Conference on Computer Vision,2019.
［11］REN S,HE K,GIRSHICK R,and et al.Towards real-time object detection with region proposal networks［C］//in Advances in Neural Information Processing Systems,2015:91-99.
［12］GIRSHICK R.Fast R-CNN［C］//The IEEE International Conference on Computer Vision,2015.
［13］GIRSHICK R,DONAHUEJ,DARRELL T,and et al.Rich feature hierarchies for accurate object detection and semantic segmentation［C］//The IEEE conference on computer vision and pattern recognition,2014.
［14］LIN T Y,GOYAL P,GIRSHICK R B,and et al.Focal loss for dense object detection［J］.IEEE Transactions on Pattern Analysis and Machine Intelligence，2020,42(2):318-327.

相似文献/References:

[1]田枫,白欣宇,刘芳,等.1种基于视频的油田危险区域入侵检测智能综合识别技术研究*[J].中国安全生产科学技术,2022,18(3):68.[doi:10.11731/j.issn.1673-193x.2022.03.010]
　TIAN Feng,BAI Xinyu,LIU Fang,et al.Research on intelligent comprehensive recognition technology of intrusion detection in oilfield dangerous area based on video[J].JOURNAL OF SAFETY SCIENCE AND TECHNOLOGY,2022,18(1):68.[doi:10.11731/j.issn.1673-193x.2022.03.010]
[2]谢谦,董立红,吴雪菲.基于多源数据融合的煤矿工作面瓦斯浓度预测*[J].中国安全生产科学技术,2022,18(11):71.[doi:10.11731/j.issn.1673-193x.2022.11.010]
　XIE Qian,DONG Lihong,WU Xuefei.Prediction of gas concentration in coal mining face based on multi-source data fusion[J].JOURNAL OF SAFETY SCIENCE AND TECHNOLOGY,2022,18(1):71.[doi:10.11731/j.issn.1673-193x.2022.11.010]
[3]崔铁军,郭大龙.基于改进YOLOX的变电站工人防护设备检测研究*[J].中国安全生产科学技术,2023,19(4):201.[doi:10.11731/j.issn.1673-193x.2023.04.029]
　CUI Tiejun,GUO Dalong.Research on detection of protection equipment for substation workers based on improved YOLOX[J].JOURNAL OF SAFETY SCIENCE AND TECHNOLOGY,2023,19(1):201.[doi:10.11731/j.issn.1673-193x.2023.04.029]
[4]曹梅,杨超宇.基于小波的CNN-LSTM-Attention瓦斯预测模型研究*[J].中国安全生产科学技术,2023,19(9):69.[doi:10.11731/j.issn.1673-193x.2023.09.010]
　CAO Mei,YANG Chaoyu.Research on CNN-LSTM-Attention gas prediction model based on wavelet[J].JOURNAL OF SAFETY SCIENCE AND TECHNOLOGY,2023,19(1):69.[doi:10.11731/j.issn.1673-193x.2023.09.010]
[5]赵泽华,王亚超,赵江平,等.基于改进YOLOv7-x的多场景火灾识别算法*[J].中国安全生产科学技术,2023,19(12):115.[doi:10.11731/j.issn.1673-193x.2023.12.015]
　ZHAO Zehua,WANG Yachao,ZHAO Jiangping,et al.Multi-scene fire recognition algorithm based on improved YOLOv7-x[J].JOURNAL OF SAFETY SCIENCE AND TECHNOLOGY,2023,19(1):115.[doi:10.11731/j.issn.1673-193x.2023.12.015]
[6]蔡晨晖,梁晓刚,师剑雄,等.矿山视频大数据智能分析与安全生产监控平台研究*[J].中国安全生产科学技术,2024,20(1):65.[doi:10.11731/j.issn.1673-193x.2024.01.010]
　CAI Chenhui,LIANG Xiaogang,SHI Jianxiong,et al.Research on mine video big data intelligent analysis and work safety monitoring platform[J].JOURNAL OF SAFETY SCIENCE AND TECHNOLOGY,2024,20(1):65.[doi:10.11731/j.issn.1673-193x.2024.01.010]
[7]王文标,时启衡,郝友维.基于改进SqueezeNet的火焰识别算法*[J].中国安全生产科学技术,2024,20(8):19.[doi:10.11731/j.issn.1673-193x.2024.08.003]
　WANG Wenbiao,SHI Qiheng,HAO Youwei.Flame recognition algorithm based on improved SqueezeNet[J].JOURNAL OF SAFETY SCIENCE AND TECHNOLOGY,2024,20(1):19.[doi:10.11731/j.issn.1673-193x.2024.08.003]
[8]韩飞腾,刘永强,房玉东,等.基于注意力机制的安全帽佩戴状态检测模型*[J].中国安全生产科学技术,2024,20(8):196.[doi:10.11731/j.issn.1673-193x.2024.08.026]
　HAN Feiteng,LIU Yongqiang,FANG Yudong,et al.Detection model for wearing status of safety helmet based on attention mechanism[J].JOURNAL OF SAFETY SCIENCE AND TECHNOLOGY,2024,20(1):196.[doi:10.11731/j.issn.1673-193x.2024.08.026]

备注/Memo

备注/Memo:
收稿日期： 2024-07-08
* 基金项目： 国家自然科学基金项目(62127808)；中国石油天然气集团有限公司科技项目（2023DJ6508）
作者简介： 魏振强，硕士，高级工程师，主要研究方向为石油石化生产过程工艺安全、工控安全及设备安全等。
通信作者： 曹欣宜，博士，高级工程师，主要研究方向为石油石化复杂系统工艺安全及控制与优化。
更新日期/Last Update: 2025-01-26