BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Long-term tracking of human pose in videos - Dr James Charles\, Re
 search Fellow at the University of Leeds: Computer Vision - Machine Learni
 ng - Probabilistic modeling
DTSTART:20160210T110000Z
DTEND:20160210T120000Z
UID:TALK64436@talks.cam.ac.uk
CONTACT:45685
DESCRIPTION:This talk presents random forest and convolutional neural netw
 ork (ConvNet) based methods for real-time human pose estimation in videos.
  I will show the proposed methods accurately track the 2D position of uppe
 r body joints\, such as the wrists and elbows\, despite fast motion and co
 ntinuously changing cluttered background. The main focus of this talk is t
 he introduction of my recently developed personalised ConvNet pose estimat
 ion method\, which greatly outperforms the current state-of-the-art on num
 erous video datasets. Furthermore\, I will introduce a new challenging vid
 eo pose dataset collected from YouTube\, and show how we can generate copi
 ous amounts of labelled pose training data efficiently and fully automatic
 ally.
LOCATION:Cambridge University Engineering Department\, Oatley Meeting Room
  (Room No. BN2-05)
END:VEVENT
END:VCALENDAR