Universite de Bordeaux I (UB1)
UB1 is one of the eight Higher Education Institutions (HEIs) composing Bordeaux University. It specializes in the fields of Mathematics, Informatics, Physics, Chemistry and the Life Sciences. UB1 is ranked top in its fields of expertise. Its cutting-edge research activities are carried out in 35 research departments, almost all of which working in close association with research performing organisations such as CNRS, INRA and INRIA. 1.000 permanent researchers along with 1.000 administrative and technical staff contribute to build UB1‘s international reputation for excellence. UB1 maintains close ties with its social environment not only through professional curriculae but also thanks to long-term private sector partnerships. Moreover it is an active partner in a number of networks and currently in two Research Training Networks, 7 ECW programmes, and one Tempus project.
Within UB1 two groups are involved in the Dem@Care project: the Video Analysis and Indexing Group of LaBRI and the Signal and Image Processing Group of IMS. The Video Analysis and Indexing Group in LaBRI is working on research related to multimedia indexing, for the analysis and interpretation of video and images, video segmentation, motion segmentation and analysis, rough indexing. It has been involved in many national and international projects, such as IP X-MEDIA or the COST292 European action. LaBRI is currently leader of the national ANR project IMMED, which involves also IMS Bordeaux, IRIT Toulouse and medical researchers of CHU Bordeaux, where wearable cameras are used for indexing the behaviour of monitored elderly people in the context of dementia diagnosis. This work follows a preliminary national PEPS S2TI CNRS project in 2007 on the same topic. These developments will serve as a basis to be extended to multiple sensors within Dem@Care. The Signal and Image Processing Group in IMS is working on research on signal processing, image processing and analysis, and their application to video. IMS is currently working on research problems involving fusion of various types of video data, processing inertial data for localisation and position estimation within national projects (PLUS, Interloc, joint project with Sagem). The Group has been involved in numerous national projects and several European projects, such as Intereg PIMHAI, Tempus CEEIM, and took an active part in the IMMED project.
In Dem@Care, UB1 leads WP4 - Situational Analysis of Daily Activities and focuses on context-aware audio-visual sensor analysis, specifically to extract meaningful information from raw sensor data and provide intermediate descriptors suitable for behaviour interpretation. Such information comes from both visual indexing of wearable videos, and fusing multiple sensor data, in order to provide more robust and precise measurements for localisation and instrumental activities.