iMCD lifelogging data : observed faces and persons

UBDC staff developed methods to analyse lifelogging image data collected within the Integrated Multimedia City Data (iMCD) project to study people's isolation level in the city.

To this end, they use the image set as an attempt to group people with similar social state and behaviour based on the number of faces and number of people they are surrounded by in their everyday life.

Using Computer Vision techniques, UBDC staff detected the number of people and the number of faces on each of the participants’ images. The result of the process was a set of features/characteristics for each participant, among which were the percentage of images they had with nobody, 1 person, etc and also with 1 face, 2 faces, etc. The next step was to cluster the participants based on the above feature space, namely create groups of people with similar social activity. After validation and experimentation, there were three clusters (groups) of people with similar social behaviour: a) low social activity, b) medium social activity, c) high social activity.

The result of this process was to enrich UBDC's existing data of participants with an additional feature, their social group, based on additional survey data.

##Access and Restrictions##
Data available for non-commercial academic research under UBDC’s End User Licence Agreement. Permitted to use for teaching purposes.