首页 | 本学科首页   官方微博 | 高级检索  
     

基于计算机视觉的预制墙板临时支撑安全合规性检查
引用本文:孙亚康,郭红领,罗柱邦,张智慧.基于计算机视觉的预制墙板临时支撑安全合规性检查[J].工程管理学报,2022,36(3):97-101.
作者姓名:孙亚康  郭红领  罗柱邦  张智慧
作者单位:清华大学 建设管理系
摘    要:预制构件安装过程中临时支撑的安全合规性对装配式施工安全至关重要,对其进行自动检查并辅助施工安全监控具有重要意义。以最为常见且具有代表性的预制混凝土墙板为例,利用 YOLO-v5 网络构建了预制墙板及其临时支撑的自动识别模型,进而提出了预制墙板临时支撑的安全合规性自动检查方法。测试结果表明,该识别模型的平均精度均值达到 95.3%,在充分考虑施工现场不利因素情况下检查方法的准确率达到 70.7%,证实了该方法的有效性和可行性。 这不仅可以提升预制墙板安装的安全性,而且还可为其他类型预制构件安装的合规性检查提供参考。

关 键 词:预制墙板  临时支撑  安全合规性检查  计算机视觉

Computer Vision-Based Safety Compliance Checking for theTemporary Support of Precast Walls
SUN Ya-kang,GUO Hong-ling,LUO Zhu-bang,ZHANG Zhi-hui.Computer Vision-Based Safety Compliance Checking for theTemporary Support of Precast Walls[J].Journal of Engineering Management,2022,36(3):97-101.
Authors:SUN Ya-kang  GUO Hong-ling  LUO Zhu-bang  ZHANG Zhi-hui
Affiliation:Department of Construction Management, Tsinghua University
Abstract:The safety compliance of temporary supports during the erection of prefabricated components is of importance toprefabricated construction, it’s quite meaningful to carry out automatic checking of the compliance and assist construction safetymonitoring. Taking commonly-seen and representative precast concrete walls as an example, this research establishes an objectdetection model for precast walls and relevant temporary supports based on YOLO-v5 network, and further proposes an automatedsafety compliance checking method for the temporary support of precast walls. It is shown from a test that the mean AveragePrecision( mAP) of the detection model is up to 95.3%, and the accuracy of the checking method is 70.7% under a harsh constructionenvironment, showing the validity and feasibility of the proposed method. This research, thus, not only improve the safety level ofthe erection of precast walls, but also provide a reference for the safety compliance checking of other precast components.
Keywords:precast wall  temporary support  safety compliance checking  computer vision
点击此处可从《工程管理学报》浏览原始摘要信息
点击此处可从《工程管理学报》下载全文
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号