在线目标跟踪系统研究与设计
首发时间:2021-04-22
摘要:目标跟踪是计算机视觉领域用于视频内容分析的重要技术手段之一。本文尝试在服务器部署MDNet目标跟踪算法,构建一个在线的目标跟踪系统,既能便于用户直观感受到目标跟踪的过程,也为后续的更深层次应用打下基础。本文设计了WEB端界面,用户在此界面上可以上传跟踪文件,并且查询跟踪结果。利用Flask后端开发框架,设计了跟踪文件传输以及结果返回的接口。MDNet模型部署在服务器上,在跟踪完成后,模型生成关于此跟踪文件的json结果文件。前端通过Flask框架从服务器获取此数据,并且最终在WEB端打印跟踪结果。最终实验表明,本文所设计的在线目标跟踪系统能有效对物体进行跟踪。
For information in English, please click here
Research and Design of Online Object Tracking System
Abstract:Object tracking is one of the important technical means for video content analysis in the field of computer vision. This thesis attempts to deploy an object tracking algorithm on the server and build an online object tracking system, which not only allows users to intuitively feel the process of object tracking, but also lays the foundation for subsequent deeper applications. This thesis designs a web page where users can upload tracking files and query the tracking results. Based on the Flask back-end development framework, an interface is designed to transfer tracking files and return results. The MDNet model is deployed on the server. After the tracking, this model generates a json result file about the tracking file. The web page fetchs this file from the server through the Flask framework, and finally displays the tracking videos. The final experiment shows that the online object tracking system designed in this thesis can effectively track objects.
Keywords: computer application technology object tracking front-end development back-end framework
基金:
引用
No.****
同行评议
勘误表
在线目标跟踪系统研究与设计
评论
全部评论