« View All Resources

KSP: Multiple Object Tracker Using K-Shortest Paths

Computer Vision Central - Posted on December 27, 2011 at 11:43 pm.

  • Links: http://cvlab.epfl.ch/software/ksp/
  • Details:
  • This code implements a multiple object tracker based on the k-shortest paths algorithm. Its input consists in a set of probabilistic occupancy maps, that is, for every time frame, a set of occupancy probabilities, one for each of the potential target locations. Such input data is usually produced by an object detector.

    The KSP object tracking algorithm is able to track an unknown and varying number of objects. Unlike other related methods, it operates on the full set of potential target locations, and not just on the detections themselves. This characteristic allows it to handle missing detections and false positives well.

    Data Sets

    Some of the multi-camera video sequences, that we acquired to test our people detection and tracking algorithms are available for download.


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

    J. Berclaz, F. Fleuret, E. Türetken and P. Fua, Multiple Object Tracking using K-Shortest Paths Optimization, accepted to IEEE Transactions on Pattern Analysis and Machine Intelligence, 2011.


    The source code is available upon request for academic purposes only and is distributed under a proprietary non-commercial license. If you are interested in using this algorithm in a commercial product, you can contact us to purchase a commercial license.


    Please mail pom@epfl.ch for bug reports, comments and questions.

    Source code

    The code is available upon request. Please contact pom@epfl.ch to obtain a copy of the source code.

    k();} ?>