G-Research & The Alan Turing Institute presents: Machines That See

A lecture given by Research Director for the Alan Turing Institute and former Laboratory Director of Microsoft Research (Cambridge), Prof. Andrew Blake: The visible world is ambiguous, so estimating physical properties by machine vision relies on probabilistic methods. Prior distributions over shape can help significantly to make finding and tracking objects more robust. Learned distributions for colour and texture make the estimators even more discriminative. These ideas fit into a philosophy of vision as inference: exploring hypotheses for the contents of a scene that explain an image as fully as possible.


To embed this video into your own website, copy and paste the code below.