Section: Overall Objectives

Highlights

Pulsar is a Project team which designs cognitive vision systems for activity recognition based on sound software engineering paradigms. We have an operational platform, named SUP, for detecting and tracking mobile objects, which can be either humans or vehicles, and for recognizing their behaviours. This SUP platform is the backbone of the team experiments to implement the new algorithms proposed by the team in perception, understanding and learning. We have studied a meta-modeling approach to support the development (e.g. specification) of video understanding applications based on SUP.

This year, we have designed an efficient algorithm for detecting people in a static image based on a cascade of classifiers. We have also proposed a new algorithm for re-identification of people through a camera network. We have realized a new algorithm for the recognition of short actions and tested its performance on several benchmarking databases. We have improved a generic event recognition algorithm by handling event uncertainty at several processing levels. We have also continued original work on learning techniques such as data mining in large multimedia databases based on offline trajectory clustering. For instance, we have been able to learn frequent activities at the apartment of an elderly, in a subway station and on an airport tarmac.

We have also started two clinical trials to characterize the behaviour profile of Alzheimer patients compared to healthy older people.

Monique Thonnat was general chair of the International Conference on Computer Vision Systems (ICVS 2011).