Max Planck Institute for Dynamics and Self-Organization -- Department for Nonlinear Dynamics and Network Dynamics Group
Personal tools
Log in

BCCN Seminar

Tuesday, 21.06.2005 16 c.t.

Prototype-based Representation of Human Body Movements: Studies in Brains and Machines

by Dr. Martin Giese
from Universität Tübingen

Contact person: J. Michael Herrmann


Seminarraum Haus 2, 4. Stock (Bunsenstr.)


The efficient representation of complex human movements is an important problem for visual perception and many technical applications. Inspired by previous work on the prototype-based encoding of complex shapes, the talk will explore representations of complex human movements that are based on learned prototypical examples. First, a learning algorithm will be presented that represents complex human action sequences by linear combination of prototypical example trajectories. The method decomposes action sequences automatically into movement primitives. It works with small amounts of training data and provides an intuitive parameterization of movement styles. The method has interesting technical applications for movement synthesis (e.g. computer animation and robotics), and movement analysis (e.g. automatic estimation of skill levels in sports, and quantification of movement disorders in neurological patients). Second, a theoretical model will be presented that shows that many results on the visual recognition of human body movements can be accounted for by a physiologically inspired neural model, which is based on learned prototypical movement patterns. Evidence from psychophysical and fMRI experiments will be presented that supports the importance of learning in biological motion recognition, and which provides some insight in the underlying neural mechanisms.

back to overview