* Cristina Manfredotti (LIP6, UPMC) - MAP5-UMR 8145

Cristina Manfredotti (LIP6, UPMC)

Relations to improve motion understanding

vendredi 26 avril 2013, 11h00 - 12h00

Salle de réunion, espace Turing

Many domains in the real world are richly structured, containing
several distinct objects interacting with each other. This is the case
of many problems as, for example, multi-target tracking, activity
recognition, automatic surveillance and traffic monitoring. The common
ground of these types of problems is the necessity of recognizing and
understanding the scene, the activities that are going on, who are the
actors, their role and estimate their positions. The explicit
representation of the interconnected behaviors of agents can provide
better models for capturing key elements of the activities in the
We develop a tracking framework that takes into account interactions
between objects allowing the recognition of complex activities. In
contrast to classic approaches that consider distinct phases of
tracking and activity recognition, our framework performs these two
tasks simultaneously. In particular, we adopt a Bayesian standpoint
where the system maintains a joint distribution of the positions, the
interactions and the possible activities. This turns out to be
advantegeous, as information about the ongoing activities can be used
to improve the prediction step of the tracking, while, at the same
time, tracking information can be used for online activity
recognition. Moreover, the explicit recognition of the relationships
between interacting objects improves the understanding of their
dynamic domain. Experimental results in two different settings show
that our approach 1) decreases the error rate and improves the
identity maintenance of the positional tracking and 2) identifies the
correct activity with higher accuracy than standard approaches.
For the learning of the possible transition models to be used by this
framework we propose a multi-layer framework, called LEMAIO, which
makes use of hierarchical abstraction to learn models for activities
involving multiple interacting objects from time sequences of data
concerning the individual objects. Experiments in the sea navigation
domain yielded learned models that were then successfully applied to
activity recognition, activity simulation and multitarget tracking.