摘要
近年来电网企业各类图片数量飞速增长,传统的图片管理系统已经不能满足用户对海量图片智能化管理和精准检索的需求。为了更加高效地智能化管理和检索图片,文章设计实现了基于深度学习的电网图片智能管理系统,利用单脉冲多波段探测器(Single Shot MultiBox Detector,SSD)技术实现图片自动标签生成功能,利用VGG16提取图像特征实现以图搜图功能。通过实际应用验证了该系统能够提升电网企业管理和检索大规模图片的效率和使用体验。
In recent years,the number of various kinds of pictures of power grid enterprises is growing rapidly,and the traditional picture management system has been unable to meet the needs of users for intelligent management and accurate retrieval of massive pictures.In order to manage and retrieve pictures intelligently and efficiently,an intelligent management system of power grid pictures based on deep learning is designed and implemented in this paper.SSD technology is used to realize automatic label generation of pictures,and VGG16 is used to extract image features to realize image retrieval.The practical application shows that the system can improve the efficiency and user experience of power grid enterprise in managing and retrieving large-scale images.
作者
闫龙川
袁孝宇
朱京
马丽
陈智雨
YAN Longchuan;YUAN Xiaoyu;ZHU Jing;MA Li;CHEN Zhiyu(State Grid Information&Telecommunication Branch,Beijing 100761,China;Beijing Kediantuoyu Intelligent Technology Co.,Ltd.,Beijing 100054,China)
出处
《电力信息与通信技术》
2021年第10期65-71,共7页
Electric Power Information and Communication Technology
基金
国家电网有限公司信息通信分公司科技项目“电力信息通信智慧运行关键场景技术研究与应用”(52993919000G)。
关键词
深度学习
图片管理
标签生成
以图搜图
deep learning
image management
label generation
image retrieval
