« View All Resources

Utrecht Multi-Person Motion Benchmark

- Posted on January 1, 2012 at 2:52 pm.

  • Links: http://www.projects.science.uu.nl/umpm/
  • Details:
  • The UMPM Benchmark is a collection of video recordings together with a ground truth based on motion capture data. It is intended to be used for assessing the quality of methods for recognition of poses from multiple persons from video data, both using a single or multiple cameras.

    The data was recorded by the Multimedia and Geometry group of Utrecht University. There are several publications that describe the content, use and recording of the benchmark. A detailed description of the packaging of the files and of the formats used, can be obtained from the data description page.

    The license conditions are mentioned on the download page.

    The authors of the benchmark can be contacted at umpm@science.uu.nl.

    Conference Paper

    paper describing the UMPM benchmark titled Utrecht Multi-Person Motion (UMPM) benchmark: a multi-person dataset with synchronized video and motion capture data for evaluation of articulated human motion and interaction and authored by N.P. van der Aa, X. Luo, G.-J. Giezeman, R.T. Tan, and R.C. Veltkamp was presented at HICV2011.

    @inproceedings{HICV11:UMPM,
      author = {Aa, N.P.~van~der and Luo, X. and Giezeman, G.J. and Tan, R.T. and Veltkamp, R.C.},
      year = 2011,
      title = {Utrecht Multi-Person Motion (UMPM) benchmark: a multi-person dataset with
        synchronized video and motion capture data for evaluation of articulated
        human motion and interaction},
      booktitle= { Proceedings of the Workshop on Human Interaction in Computer Vision (HICV), in
          conjunction with ICCV 2011}
    }
    

    Presentation

    zip file with the HICV powerpoint presentation and several videos is also available. The videos can give a quick impression of the kind of scenarios and the data that are available.

    Technical Report

    technical report describes details of the benchmark and the recording process.


    k();} ?>