基于BIM和YOLOv5的闸墩浇筑高度智能识别方法  

Intelligent identification method of pouring height of sluice pier based on BIM and YOLOv5

作　　者：刘奕炜 陈铭轩 牛志伟[1] 丁毅 陈荣 LIU Yiwei;CHEN Mingxuan;NIU Zhiwei;DING Yi;CHEN Rong(College of Water Resources and Hydropower,Hohai University,Nanjing 210098,China;Nanjing Hongyuan Information Technology Co.,Ltd.,Nanjing 210000,China;Yizheng Municipal Water Conservancy Bureau,Yangzhou 225000,China)

机构地区：[1]河海大学水利水电学院,江苏南京210098 [2]南京鸿源信息技术有限公司,江苏南京210000 [3]江苏仪征市水利局,江苏扬州225000

出　　处：《人民长江》2024年第9期238-243,共6页Yangtze River

基　　金：江苏省水利科技项目(2022080);国家自然科学基金青年基金项目(52009035)。

摘　　要：水利工程施工进度是工程建设方关注的重要事项。为实现水闸工程的智慧化建设,基于BIM技术和图像识别技术提出了一种虚实结合的闸墩浇筑高度识别方法。通过BIM技术建模,生成了包含不同闸墩浇筑高度的虚拟图像数据集;利用深度学习的目标检测算法YOLOv5对虚拟数据集进行训练,得到能准确识别不同浇筑高度闸墩的智能模型。在实验室搭建具有不同浇筑高度的闸墩模型,通过摄像机拍摄获取不同高度闸墩的实景图像,再利用该智能识别模型对实景图像进行验证。研究结果表明,该模型能够准确识别实际场景中闸墩的浇筑高度,可广泛应用于水利工程主体建筑物的实际施工进度识别,为数字孪生水利工程建设提供有效的模型支持。The construction schedule of water conservancy projects is an important matter of great concern to the construction parties.Based on BIM technology and image recognition technology,this paper proposed a virtual and real combining method for pouring height recognition of sluice piers,so as to realize the intelligent construction of sluice projects.Through BIM technology modeling,virtual image data sets containing different pouring heights of gate piers were generated.The object detection algorithm YOLOv5 of deep learning was used to train the virtual data set,and an intelligent model of accurately identifying piers with different pouring heights was obtained.The real pier models with different pouring heights were built in the laboratory,and the real images of the piers with different heights were captured by the camera and verified by the established intelligent recognition model.The results showed that the model can accurately identified the pouring heights of piers in actual scenes.This method can be widely used to identify the actual construction progress of buildings in hydraulic engineering,and provide effective model support for the construction of digital twin hydraulic engineering.

关 键 词：水闸闸墩 BIM YOLOv5 计算机视觉技术 深度学习 

分 类 号：TV511[水利工程—水利水电工程]