Python wrapper for “Multiple Object Tracker Using K-Shortest Paths”

Wrapper written by Pierre Baqué from the original "Multiple Object Tracker Using K-Shortest Paths" code.

Usage

If you want to modify the source code or need recompile the wrapper, enter the command:

cd pyKSP
python setup.py build_ext --inplace

Or install to your site-packages to make the module available everywhere:

cd pyKSP
python setup.py install

The python script in "pyKSP-Example.py", provides an example on how to use pyKSP from Python.

Dependencies

Python libraries:

pip install cython numpy matplotlib

Boost Graph Library to build the KSP binary:

# on ubuntu
sudo apt install libboost-graph-dev

References

For more information about the KSP algorithm, please check the following article:

Jerome Berclaz, Francois Fleuret, Engin Turetken and Pascal Fua, "Multiple Object Tracking using K-Shortest Paths Optimization", IEEE Transactions on Pattern Analysis and Machine Intelligence (TPAMI), 2011.