BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Machines that See\, Powered by Probability - Professor Andrew Blak
 e\, Microsoft Research Cambridge
DTSTART:20131204T160000Z
DTEND:20131204T170000Z
UID:TALK49027@talks.cam.ac.uk
CONTACT:CCA
DESCRIPTION:Machines with some kind of ability to see have become a realit
 y in the last decade\, and we see vision capabilities in cameras and photo
 graphy\, cars\, graphics software and in the user interfaces to appliances
 . Such machines bring benefits to safety\, consumer experiences\, and heal
 thcare. \n\nThe visible world is inherently ambiguous and uncertain\, so e
 stimation of physical properties by means of vision tends to rely on proba
 bilistic methods. Introducing regularization into functionals used in opti
 mal estimation already helps absorb noise in sensory data\, but visual pro
 cessing makes further demands on the mechanisms of probability. Prior dist
 ributions over shape can help signficantly to make estimators of shape mor
 e robust. Learned distributions for colour and texture are used to make es
 timators more discriminative. These ideas support inference by finding hyp
 otheses for the contents of a scene that explain an image as fully as poss
 ible. More recently this explanatory approach has somewhat  given way to p
 owerful direct estimation methods\, with parameters tuned using large trai
 ning sets. Perhaps the most capable vision systems will come ultimately fr
 om some kind of fusion of the two approaches.\n
LOCATION:MR2
END:VEVENT
END:VCALENDAR