|Description (include details on usage, files and paper references)||The High Definition Analytics (HDA) dataset is a multi-camera High-Resolution image sequence dataset for research on High-Definition surveillance: Pedestrian Detection (PD), person Re-Identification (RE-ID), fully integrated PD and RE-ID (PD+REID), inter-camera tracking.
18 cameras (including VGA, HD and Full HD resolution) were used to acquire video streams simultaneously during 30 minutes in a typical indoor office scenario at a busy hour (lunch time) involving more than 80 persons. In the current release (v1.2), 13 of the video streams have been fully labelled.
The dataset includes:
* image sequences,
* annotations (Bounding Boxes with the person ID and a visibility flag),
* a default gallery and set of probes (complete with body part detections) for benchmarking classical Re-Identification algorithms,
* detections obtained by a PD algorithm (complete with body part detections), for evaluating a RE-ID algorithm in a PD+REID schema without having to implement a PD algorithm,
* and, finally, standard code for evaluating PD, RE-ID, PD+REID and inter-camera tracking algorithms.
We maintain a public repository of RE-ID results on this dataset at:
"A Multi-camera video data set for research on High-Definition surveillance", Int. J. of Machine Intelligence and Sensory Signal Processing, 2014.
"The HDA+ data set for research on fully automated re-identification systems", ECCV workshop, 2014.