Dem@Care Project – CHU Dataset
Dem@Care is providing a public dataset, which is collected during lab experiments. The data collection took place in Centre Hospitalier Universitaire de Nice (CHU) in Nice, France. The datasets include video and audio recordings as well as data from physiological sensors. However, due to privacy regulations, only tracking information is available to the research community.
General Info:
– The people recruited for this dataset were 27 elderly people, aged 65 and above, of both genders.
– The participant pool included people that were healthy and people suffering from conditions ranging from Mild Cognitive Impairment (MCI) to mild dementia, and in a few cases full-blown Alzheimer’s Disease (AD).
– The participants are asked to perform following activities of daily living (ADL):
Activities of Interest:
– Reading Article
– Watering Plant
– Preparing Drug Box
– Preparing Drink
– Watching TV
– Talking On Phone
– Checking Bus Map
Tracking Data:
– Videos have been processed for people detection and tracking using [1] and [2].
– Tracking data for 27 videos can be downloaded from here: Tracking_Data_CHU
– Tracking data contains following information in XML-format:
i) Timestamp
<MobFrame timeYear=“2012” timeDay=“25” timeMonth=“6” timeMs=“949” timeMin=“40” frameID=“14” timeHour=“14” timeSec=“41” nbMobiles=“1”>
ii) 2D information (please ignore speed information):
<Info2D ySpeed=“0” timeDay=“25” timeHour=“14” yCenter=“114” xSpeed=“0” timeMonth=“6” timeSec=“41” timeMin=“40” yCog=“114” timeYear=“2012” frameID=“14” width=“169” nbMovPix=“0” orientation=“0” xCog=“515” xCenter=“515” camID=“0” length=“68” timeMs=“949” />
iii) 3D information (please ignore speed information):
<Info3D w3D=“662” zSpeed=“0” x=“2028.08” ySpeed=“0” y=“-765.066” h3D=“1561.11” xSpeed=“0” z=“5499.51” l3D=“1097” orientation=“0” />
Ground Truth:
– Activities are manually annotated for each video. The ground truth can be downloaded from here: Ground_Truth_CHU
Calibration Data:
– 3D points can be projected on the image plane using the intrinsic matrix:
[ 594.2 0. 339.3;
0. 591. 242.7;
0. 0. 1. ]
[1] Anh-Tuan Nghiem, Edouard Auvinet, and Jean Meunier. Head detection using kinect camera and its application to fall detection. In 11th International Conference on Information Science, Signal Processing and their Applications, ISSPA, pages 164–169, 2012.
[2] Duc Phu Chau, Julien Badie, François Bremond, and Monique Thonnat. Online Tracking Parameter Adaptation based on Evaluation. In IEEE International Conference on Advanced Video and Signal-based Surveillance, Krakow, Poland, August 2013.
