CSR-DCF视频目标跟踪源码运行笔记
这篇论文 titled "Discriminative Correlation Filter with Channel and Spatial Reliability" was authored by Alan Lukezic and colleagues and published in CVPR 2017. The research introduces a novel correlation filter method that incorporates both channel and spatial reliability for improved object tracking. The paper provides detailed information about the experimental setup, including the required running environment, which involves Windows 10 x64, MATLAB R2016b, Visual Studio 2013, and OpenCV 2.4.12. Key steps include installing OpenCV with its path correctly configured in the system variables, setting up the MATLAB C/C++ compiler, and running the provided demo script after modifying the relevant paths and video names. The authors emphasize the importance of using compatible versions of MATLAB, Visual Studio, and OpenCV to ensure successful implementation. Additionally, the paper encourages readers to follow the "视觉边疆"微信订阅号以获取更多文章推送。
论文基本信息
论文标题:基于通道和空间可靠性差异的判别相关滤波器
论文作者:Alan Lukezic和others
论文出处:Proceedings of the Conference on Computer Vision and Pattern Recognition (CVPR) 2017
文章链接:https://arxiv.org/abs/1611.08461
补充材料:https://www.semanticscholar.org/paper/Discriminative-Correlation-Filter-with-Channel-and-Lukežič-Vojı-ř/7b485979c75b46d8c194868c0e70890f4a0f0ede
源码链接:https://github.com/alanlukezic/csr-dcf
运行环境介绍
- Windows 10 64-bit
- 建议使用MATLAB R2016版本,因为2017及以后的版本可能存在问题
- Visual Studio 2013版本
- OpenCV 2.4.12版本,可通过访问https://opencv.org/releases.html获取
注意:建议选用相同版本的软件组合,其他组合可能导致问题
准备
安装OpenCV库,并确保将OpenCV的路径正确配置到系统变量的Path变量中。在安装程序默认未配置的情况下,建议手动添加。如图所示:

在MATLAB环境中,通过配置Visual Studio 2013与C及C++编译器,可以执行以下命令来查看配置信息。建议在命令行界面中分别运行mex -setup和mbuild命令,以完成配置验证。
mex -setup
mex -setup C++
如果默认不是VS 2013的编译器,则根据MATLAB的提示进行操作即可。
运行
在MATLAB环境中解压CSR-DCF的源码文件后,进入指定路径并打开名为compile.m的文件。随后,需要修改第15和17行的代码部分,将路径配置为与自己安装的OpenCV相关的路径。特别提示:确保路径末尾添加反斜杠符号,如图所示。

运行compile.m文件,此时会得到mex完成的输出,如下图所示:

在终端中输入./demo_csr.m,根据实际需求,修改文件中的两处路径信息和视频文件名称,确保修改后不再留有反斜杠符号,如图所示。

运行demo_csr.m文件,效果如下图所示:

想持续获得文章推送?欢迎扫码关注“视觉边疆”微信订阅号

