基于深度学习的隧道渗漏水语义分割方法
Segmentation Method of Leak in Tunnel Based on Deep Learning
徐艺文 1王维 1王鲁杰 2陈颖 2郭春生 1李家平2
作者信息
- 1. 上海勘察设计研究院(集团)股份有限公司
- 2. 上海地铁监护管理有限公司
- 折叠
摘要
文章针对隧道结构渗漏水病害巡检效率低的问题,基于隧道结构三维激光扫描影像建立了一个具备了一定规模的渗漏水病害数据集,选择了三种经典的图像分割的深度学习模型,分析和比较了三种模型在渗漏水病害识别的区别及差异,验证了图像分割的深度学习模型的有效性.
Abstract
Aiming at the problem of low efficiency in the inspection of water leakage diseases in tunnel structures,this paper establishes a data set of water leakage diseases with a certain scale based on 3D laser scanning images of tunnel structures,and chooses three classic deep learning models for image segmentation,analyzes and compares the distinctions and differences of the three models in the recognition of water leak disease,and verifies the effectiveness of the deep learning model for image segmentation.
关键词
隧道三维激光扫描影像/深度学习/数据集/渗漏水Key words
tunnel image of 3D laser scan/deep learning/dataset/leakage引用本文复制引用
出版年
2024
NETL
NSTL
维普
万方数据