商超自动货架巡检系统设计

Design of an Automatic Shelf Inspection System

文章设计了基于物联网、机器视觉和深度学习技术的自动货架巡检系统。系统包涵数据采集、数据分析以及数据统计等三个方面的功能,系统设计由终端和数据中心两个部分组成。终端包括巡检机器人、WEB终端以及手机APP,数据中心由业务平台、深度学习平台和大数据平台组成。其中业务平台基于容器/虚拟化技术,为每个用户提供网络隔离、数据隔离、应用隔离的操作系统环境。基于该环境,向用户发布若干应用模块:深度学习平台基于采集的数据,进行迭代离线训练,输出预测器,向业务系统发布;大数据平台用于托管用户数据,并进行大数据分析。

This paper designs an automatic shelf inspection system based on Internet of things,machine vi⁃sion and deep learning technology.The system includes data acquisition,data analysis and data statistics.The system design is composed of terminal and data center.The terminal includes inspection robot,Web terminal and mobile app.The data center is composed of business platform,deep learning platform and big data platform.The business platform is based on container/virtualization technology,providing each user with an operating system environment of network isolation,data isolation and application isolation.Based on the environment,several ap⁃plication modules are released to users;the deep learning platform conducts iterative offline training based on the collected data,outputs predictors and releases them to business systems;big data platform is used to host user da⁃ta and conduct big data analysis.