基于深度学习技术的“云共治”架构设计  

Design of"Cloud Governance"Architecture Based on Deep Learning Technology

作　　者：吴俊龙 牟建建 王震 马玉亮 WU Jun-long;MOU Jian-jian;WANG Zhen;MA Yu-liang(Hangzhou Qiantang Wasu Digital TV Co.,Ltd.,Hangzhou,Zhejiang 311225,China;Wasu Media Network Co.,Ltd.,Hangzhou,Zhejiang 310051,China)

机构地区：[1]杭州钱塘华数数字电视有限公司,浙江杭州311225 [2]华数传媒网络有限公司,浙江杭州310051

出　　处：《电子技术与软件工程》2024年第4期26-34,共9页ELECTRONIC TECHNOLOGY & SOFTWARE ENGINEERING

摘　　要：随着城市化的发展,城市基层治理难度不断加大,大量影响群众安全和生活的治理事件长期存在也难以根除,以现有的传统手段去治理已远远不够。本文研究提出了一种基于卷积神经网络深度学习技术的智能识别系统,并构建了违法主体、基层社区、执法机关共同治理形成闭环逻辑圈的“云共治”架构设计方案,对城市管理中的违规行为进行有效感知、识别与分派,构建了基于视频数据分析推进城市治理高效化的新模式。方案采用GB/T 28181—2022标准组建了视频汇聚和级联系统,解决了异构监控系统视频资源共享的技术问题,实现了大规模数据之间的互通共享,提升了系统集约化建设水平。本方案利用大数据技术,创建大数据研判模型,实现城市基层治理事件的准确分类及派发,具备二次分析应用能力。实验结果表明,该系统综合识别率达到90%以上。With the development of urbanization,the difficulty of urban grassroots governance continues to increase,and a large number of governance events that affect people's safety and life are difficult to eradicate for a long time.It is far from enough to govern with existing traditional means.This paper studies and proposes an intelligent identification system based on the"convolutional neural network"deep learning technology,and constructs a"cloud co governance"architecture design scheme for the joint governance of illegal subjects,grass-roots communities and law enforcement agencies to form a closed-loop logical circle.It effectively senses,identifies and assigns violations in urban management,and constructs a new mode to promote the efficiency of urban governance based on video data analysis.The scheme uses the GB/T 28181-2022 standard to build a video convergence and cascade system,which solves the technical problem of video resource sharing in heterogeneous monitoring systems,realizes the interoperability and sharing of large-scale data,and improves the level of system intensive construction.The scheme uses big data technology to create a big data research and judgment model to realize the accurate classification and distribution of urban grass-roots governance events,and has the ability of secondary analysis and application.The experimental results show that the comprehensive recognition rate of the system has reached more than 90%.

关 键 词：深度学习 云共治 大数据分析 视频接入整合 事件识别 基层治理 

分 类 号：F299.24[经济管理—国民经济] TP391.4[自动化与计算机技术—计算机应用技术]