Body Laughter Index
Posted on Wed 06 Mar 2013 by Maurizio Mancini
Step 1. In the first video a manual tracking of head
and shoulders is carried out: the cloud of red points determines the
Regions Of Interest (ROIs) head and shoulders are located in. The blue
dots are the geometrical barycenters of each cloud. In the second video, shoulders are automatically tracked by means of small
green markers.
Step 2. Body Laughter Index is computed by looking at shoulders' correlation and energy: high correlation and energy indicate that the probability of laughter is high; this information is acknowledged only if the shoulders' movement is periodic.
Step 2. Body Laughter Index is computed by looking at shoulders' correlation and energy: high correlation and energy indicate that the probability of laughter is high; this information is acknowledged only if the shoulders' movement is periodic.
Greta laughter
Posted on Tue 05 Mar 2013
Results of motion retargeting by learning a mapping between motion capture data and facial animation parameters of a virtual agent.
CNRS - TELECOM ParisTech
Abhishek Sharma, Radoslaw Niewiadomski, Catherine Pelachaud
CNRS - TELECOM ParisTech
Abhishek Sharma, Radoslaw Niewiadomski, Catherine Pelachaud
Real-Time Laughter in Motion Builder
Posted on Mon 04 Feb 2013 by Husyein Cakmak
Using the Optitrack face motion capture system, virtual characters in
Motion Builder can be animated in real-time. See the example in the
video below, which includes some laughs. This method is appealing for
Wizard-of-Oz systems, but the ILHAIRE Consortium is currently
investigating ways to automatise real-time audiovisual synthesis of such
laughter sequences, without the need for a performer to drive the agent.