Multi Object Tracking using OpenCV

Instructions for C++ code

Download the code and extract the downloaded file, if required.
Create a build folder: mkdir build && cd build.
System specific instructions:
1. For Windows Operating System: Tell CMake to configure and generate project files in Visual Studio 14 x64 format. Use .. to look for CMakeLists.txt file in the parent folder of build directory. It will generate project files in Visual Studio 14’s format. cmake -G "Visual Studio 14 2015 Win64" ..
2. For Linux/MacOS: Tell CMake to configure and generate project files. .. denotes that CMake should look for CMakeLists.txt file in the parent folder of build directory. For Linux/MacOS cmake detects and uses the C++ toolchain installed on your system It will most likely be gcc toolchain. cmake ..
Now tell CMake to build project in Release mode: cmake --build . --config Release
When the build is completed, executables will be generated in build\Release folder or build folder itself.
Now go back to project directory (parent directory of build) and run the compiled script: ./build/multiTracker

The scripts use CSRT as the default tracking algorithm. To use any other algorithm, change trackerType variable to the tracker of your choice.

Note that only these tracking algorithms are available in the scripts: BOOSTING, MIL, KCF, TLD, MEDIANFLOW, GOTURN, MOSSE, CSRT.

To change the trackerType variable in C++, go to line number 58 and change the tracker to "BOOSTING", for example.

To do the same for Python code, edit trackerType variable in line number 46.

For instructions regarding how to use GOTURN tracker, please refer to the blog on GOTURN: Deep Learning based Object Tracking and the corresponding GitHub repository.

Want to become an expert in AI? AI Courses by OpenCV is a great place to start.