BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Automatic Multimodal Descriptors of Rhythmic Body Movement - Marwa
 \, Jingjing
DTSTART:20131128T111500Z
DTEND:20131128T121500Z
UID:TALK48524@talks.cam.ac.uk
CONTACT:Jingjing Shen
DESCRIPTION:Prolonged durations of rhythmic body gestures were proved to b
 e correlated with diff erent types of psychological disorders. To-date\, t
 here is no automatic descriptor that can robustly detect those behaviours.
  In this talk\, we propose a cyclic gestures descriptor that can detect an
 d localise rhythmic body movements by taking advantage of both intensity a
 nd depth modalities. We show experimentally how our rhythmic descriptor ca
 n successfully localise the rhythmic gestures as: hands fidgeting\, legs f
 idgeting or rocking\, significantly higher than the majority vote classifi
 cation baseline. Our experiments also demonstrate the importance of fusing
 \nboth modalities\, with a significant increase in performance when compar
 ed to individual modalities.
LOCATION:Rainbow Seminar Room (SS03)\, Computer Laboratory
END:VEVENT
END:VCALENDAR