净水技术2025，Vol.44Issue(6)：157-163,7.DOI:10.15890/j.cnki.jsjs.2025.06.019
基于深度残差神经网络的道路积水深度提取方法
Extraction Method of Road Waterlogging Depth Based on Deep Residual Neural Network
陈汪洋1
作者信息
- 1. 中国水务投资有限公司,北京 100053
- 折叠
摘要
Abstract
[Objective]In response to the high installation and maintenance costs of existing road waterlogging depth sensors,this paper proposed a road waterlogging depth extraction method based on deep residual neural network.[Methods]The method collected data from multiple channels such as the"global eye system""network platform"and"flood emergency rescue photos",[Results]to address the issues of insufficient quantity and low quality of the waterlogging image dataset.By labeling the dataset based on the severity of waterlogging,the method solved the problem of inability to identify waterlogging depth in previous research,achieving an accuracy of 96.5％on the original test set.[Conclusion]Conclusion from case analysis demonstrate the feasibility of using deep residual neural network to extract waterlogging depth information from waterlogging images,meeting the accuracy requirements for waterlogging monitoring in practical applications.关键词
排水系统/道路积水/残差神经网络/积水深度提取/内涝防控Key words
drainage system/road waterlogging/residual neural network/waterlogging depth extraction/waterlogging control分类
建筑与水利引用本文复制引用
陈汪洋..基于深度残差神经网络的道路积水深度提取方法[J].净水技术,2025,44(6):157-163,7.
