Presentation

These pages describe a people detection evaluation challenge, named People Detection benchmark (PDbm). It includes realistic sequences extracted from [2], manually annotated people detection ground truth and a complete evaluation framework.

Following the idea in [1] for change detection. We propose a People Detection benchmark repository (PDbm) [3]. This benchmark allows to easily compare any people detection over a realistic, large-scale dataset with non-specific or constrained scenarios. In addition, the results will be available for the research community.

Main reference:

        [1] A. García-Martín, J.M. Martínez, J. Bescós,  “PDbm: People detection benchmark repository”, Electronic Letters Volume 51, Issue 7, p. 559 – 560 DOI: 10.1049/el.2014.3795 , Print ISSN 0013-5194, Online ISSN 1350-911X

Associated references:

[2] 2012 IEEE Change Detection Workshop, http://www.changedetection.net/.

[3] People Detection Benchmark http://www-vpu.eps.uam.es/PDbm/

Recall that this people detection benchmark is available just for research purposes.

Work partially supported by project TEC2011-25995 EventVideo.