期刊文献+

基于计算机视觉与深度学习的施工现场人员及车辆智能统计与监测

Construction Site Personnel and Vehicles Statistics and Monitoring Based on Computer Vision and Deep Learning
下载PDF
导出
摘要 近年来,我国的建筑业飞速发展,建筑业企业生产和经营规模的不断扩大,单个项目体量也变得越来越大,极大地增加了建筑施工现场人员管理的难度。为提高施工现场的人员管理效率,借助计算机视觉技术以及深度学习算法实现施工现场人员统计与监测,利用闭路监控的实时画面实现工地现场人员的实时、高效、自动化分析。该法构建基于YOLOv4的深度学习算法模型,在识别出画面中施工人员的同时,通过撞线判定机制统计进入与离开工地的施工人员数量。除此之外,还研究开发了面向工地管理人员的施工人员管理系统,并通过试验验证该系统在不同环境条件下对施工现场人员进出场情况实时统计与监测的准确性。在现场测试过程中,发现系统在画面检测范围、算法置信度、系统冗余程度存在关键制约性问题,通过解决上述问题,极大地提高了系统的实际应用价值和鲁棒性,为系统在实际工程项目中应用打下基础。 With the rapid development of the construction industry,large-scale projects have gradually increased,and such projects have the characteristics of large on-site personnel and complex management.In order to improve the efficiency of personnel management at the construction site,the statistics and monitoring of construction site personnel can be realized with the help of computer vision technology and deep learning algorithms.The real-time video of closed-circuit television is used to realize the real-time,efficient and automated analysis of site personnel.This method constructs a deep learning algorithm model based on YOLOv4.While identifying the construction personnel in the picture,the number of construction personnel entering and leaving the construction site is counted by means of the collision line determination mechanism.In addition,a construction personnel management system for construction site managers has also been developed.In the end,this paper verified the accuracy of real-time statistics and monitoring of the entry and exit of construction site personnel under different environmental conditions through tests.In the process of field testing,it is found that there are key constraints in the screen detection range,algorithm confidence and system redundancy degree of the system.By solving the above problems,the practical application value and robustness of the system are greatly improved,laying a foundation for the application of the system in actual engineering projects.
作者 陈祺荣 陈钰开 林俊 朱东烽 CHEN Qirong;CHEN Yukai;LIN Jun;ZHU Dongfeng(Guangdong Juncheng Construction Technology Co.,Ltd.Yunfu 527300,China;School of Civil Engineering and Architecture,Hainan University Haikou 570228,China)
出处 《广东土木与建筑》 2024年第4期10-14,共5页 Guangdong Architecture Civil Engineering
关键词 计算机视觉 深度学习 YOLOv4 施工人员统计与检测 computer vision deep learning YOLOv4 construction personnel statistics and monitoring
  • 相关文献

参考文献11

二级参考文献28

共引文献28

相关作者

内容加载中请稍等...

相关机构

内容加载中请稍等...

相关主题

内容加载中请稍等...

浏览历史

内容加载中请稍等...
;
使用帮助 返回顶部