Home »


Repository of Results


We maintain a public repository of re-identification results in this dataset. Send us your CMC curve to be uploaded  (alex isr ist utl pt).

In our ECCV 2014 paper, we describe six experiments:

  • MANUALall: Re-identification on manual annotated data;
  • MANUALclean: Re-identification on manual annotated data, but only of fully visible upright people;
  • DIRECT: Re-identification on Pedestrian Detection data computed with the ACF algorithm;
  • FP OFF, OCC ON: Re-identification on the Pedestrian Detection data, but filtering out some of the occluded samples;
  • FP ON, OCC OFF: Re-identification on the Pedestrian Detection data, but also training an extra class to represent False Positives;
  • FP ON, OCC ON: Re-identification on the Pedestrian Detection data, filtering out some occluded samples and using the extra False Positive class.

We invite you to submit your results on the HDA+ dataset and we present here the up-to-date results on the different experiments.

In the legend, the first number correponds to re-identification accuracy at rank 1, and the second number represents area under the CMC curve.

Results with test camera 60, and remaining cameras for training:

MANUAL_c_l_e_a_n cam60MANUAL_a_l_l, All camerasDIRECT (FP OFF, OCC OFF) cam60FP OFF, OCC ON cam60FP ON, OCC OFF cam60FP ON, OCC ON cam60

Experiments with cross-validation across all cameras, one camera held for test with the rest for training for each round. Results averaging the 13 rounds of cross-validation:

MANUAL_c_l_e_a_n, All camerasMANUAL_a_l_l, All camerasDIRECT (FP OFF, OCC OFF), All camerasFP OFF, OCC ON, All camerasFP ON, OCC OFF, All camerasFP ON, OCC ON, All cameras

Note: There are no results for [4] due to memory issues with the present form of the code.

Dario Figueira, Matteo Taiana, Athira Nambiar, Jacinto Nascimento, Alexandre Bernardino, ”The HDA+ data set for research on fully automated re-identification systems“, in ECCV workshop, 2014.
[2] Forssen, P.-E., “Maximally Stable Colour Regions for Recognition and Matching,” in Computer Vision and Pattern Recognition (CVPR), 2007.
[3] M. Farenzena, L. Bazzani, A. Perina, V. Murino, and M. Cristani, ”Person Re-identification by Symmetry-Driven Accumulation of Local Features”  In IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2010
[4] Dario Figueira, Loris Bazzani, Ha Quang Minh, Marco Cristani, Alexandre Bernardino, Vittorio Murino, “Semi-supervised multi-feature learning for person re-identification“, in IEEE International Conference Advanced Video and Signal Based Surveillance (AVSS), 2013.