Crowd Dataset + Ground Truth crowd detection dataset

[Back]

Download

Download Here

NOTE: The dataset is publicly available for non-commercial use. Please refer to Lim et al, ICPR, 2014 if you use this dataset in your publication.

Details

The crowd datasets are obtained a variety of sources, such as UCF and Data-driven crowd datasets to evaluate the proposed framework. The sequences are diverse, representing dense crowd in the public spaces in various scenarios such as pilgrimage, station, marathon, rallies and stadium. In addition, the sequences have different field of views, resolutions, and exhibit a multitude of motion behaviors that cover both the obvious and subtle instabilities.

There are two(2) folders associated with the dataset and a ReadMe file:

  1. Crowd Sequence
    • Folder name = Crowd Sequence
    • Total Sequences = 20
  2. Dataset Groundtruth
    • Folder name = Dataset Groundtruth
    • Total Files (text) = 20
    • The salient regions are defined by bounding box information in the form of either:
      - Rectangle Properties (Frame no; Region Identifier; Minimum x-coordinate; Minimum y-coordinate; Maximum x-coordinate; Maximum y-coordinate)
      - Example: 34 1 75 76 140 130

      - Polygon Properties (Frame no; Region Identifier; Point 1 x-coordinate; Point 1 y-coordinate; Point 2 x-coordinate; Point 2 y-coordinate; Point 3 x-coordinate; Point 3 y-coordinate; Point 4 x-coordinate; Point 4 y-coordinate)
      - Example: 1 11 153 182 323 121 342 173 170 232

Citation

Crowd Saliency Detection via Global Similarity Structure
M.K. Lim, V.J. Kok, C.C. Loy and C. S. Chan
Proceedings of 22nd International Conference on Pattern Recognition, pp. 3957-3962, 2014 (ICPR 2014)
 PDF

Acknowledgement

This research is supported by the Fundamental Research Grant Scheme (FRGS) MoE Grant FP027-2013A, H-00000- 60010-E13110 from the Ministry of Education Malaysia. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the University of Malaya.

DISCLAIMER: Electronic paper shared on this website are for fast dissemination of research work. The copyrights of the papers belong to the corresponding publishers.