BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Closing the gap between weakly and fully supervised methods - Vitt
 orio Ferrari\, ETH Zurich
DTSTART:20111201T150000Z
DTEND:20111201T160000Z
UID:TALK34859@talks.cam.ac.uk
CONTACT:Microsoft Research Cambridge Talks Admins
DESCRIPTION:Learning visual classes is traditionally done in a fully super
 vised setting\, where the location of training instances are manually anno
 tated by bounding-boxes or pixelwise segmentations. After learning a model
  of a class from this data\, it can be used to recognise and localise nove
 l instances in test images. On the other hand\, weakly supervised techniqu
 es try to learn such models from training images labeled only by the prese
 nce or absence of the class\, without location annotation. While these tec
 hniques can substantially reduce the manual effort necessary to learn a cl
 ass\, fully supervised methods typically deliver models that perform consi
 derably better on test data. In this talk I will present recent advances t
 owards closing this performance gap for the two tasks of action recognitio
 n and semantic segmentation.\nThese are steps towards the goal of devising
  weakly supervised techniques that deliver models of the same quality as f
 ully supervised ones.
LOCATION:Small public lecture room\, Microsoft Research Ltd\, 7 J J Thomso
 n Avenue (Off Madingley Road)\, Cambridge
END:VEVENT
END:VCALENDAR