BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Body Part Recognition: Making Kinect Robust - Jamie Shotton\, Micr
 osoft Research Cambridge
DTSTART:20120705T083000Z
DTEND:20120705T093000Z
UID:TALK38764@talks.cam.ac.uk
CONTACT:Microsoft Research Cambridge Talks Admins
DESCRIPTION:Microsoft recently launched Xbox Kinect (http://www.xbox.com/k
 inect)\, a revolution in gaming where your whole body becomes the controll
 er.  Human pose estimation has long been a “grand challenge” of comput
 er vision\, and Kinect has been the first product that meets the speed\, c
 ost\, accuracy\, and robustness requirements to take pose estimation out o
 f the lab and into the living room.\n\nIn this talk we will discuss some o
 f the challenges of pose estimation and the technology behind Kinect\, det
 ailing a new algorithm\, body part recognition\, which drives Kinect’s s
 keletal tracking pipeline.  Body part recognition uses a classifier to pro
 duce an interpretation of pixels coming from the Kinect depth-sensing came
 ra into different parts of the body: head\, left hand\, right knee\, etc. 
  Estimating this pixel-wise classification is extremely efficient in paral
 lel on the GPU.  The classifications are then pooled across pixels to prod
 uce hypotheses of 3D body joint positions for use by a skeletal tracking a
 lgorithm.  Our method has been designed to be robust\, in two ways in part
 icular.  First\, we train the system with a vast and highly varied trainin
 g set of synthetic images to ensure the system works for all ages\, body s
 hapes & sizes\, clothing and hair styles.  Second\, the recognition does n
 ot rely on any temporal information\, and this ensures that the system can
  initialize from arbitrary poses and prevents catastrophic loss of track\,
  enabling extended gameplay for the first time.
LOCATION:Large lecture theatre\, Microsoft Research Ltd\, 7 J J Thomson Av
 enue (Off Madingley Road)\, Cambridge
END:VEVENT
END:VCALENDAR