摘要
对于防洪河道的视频监控是水利信息化的重要内容,对此类视频智能分析,挖掘视频内在价值,对于推进智慧水务建设具有重要意义。该研究通过应用场景的需求分析,硬件电路开发,基于U-Net的水位识别算法开发与移植应用,实现了边缘侧视频无标尺水位监测。通过现场试验,分析了水位识别的准确性,可作为河道水位监测的补充手段。
Video monitoring of rivers is an important part of water conservancy informatization.Intelligent analysis of such videos and excavation of the intrinsic value of videos are of great significance for promoting the construction of smart water affairs.This study achieved edge side video scaleless water level monitoring by analyzing the requirements of application scenarios,developing hardware circuits and porting U-Net based water level recognition algorithms.The accuracy of water level recognition has been analyzed through on-site experiments,which can serve as a supplementary means for monitoring river water levels.
作者
聂明杰
NIE Mingjie(Beijing Research Institute of Smart Water,Beijing 100036,China)
出处
《北京水务》
2022年第S02期73-76,共4页
Beijing Water
基金
北京水务青年科技骨干人才培养计划资助(BWAPT202203):“水务视频边缘计算平台应用研究”。
关键词
深度学习
边缘计算
智慧水务
deep learning
edge computing
smart water platform
