Professor Philip Torr
Wolfson Research Merit Award Holder
Department of Computing, Oxford Brookes University
"In the past forty years, computers and digital cameras have increased in both power and prevalence. But how do we make computers see the way we see? Humans make sense of the movement, colours, objects and people we see in our environment, and my research is dedicated to transferring this visual interpretation to the world of computers. There has already been a large amount of research in computer vision, leading to an in-depth understanding of functions like feature detectors, tracking, camera calibration, and segmentation, but there has been only a limited amount of research on putting all this together into a framework that would allow the computer or camera to recognise the objects in the scene and be able to describe their shape.
This can be achieved by usingcomplex models that capture the spatial information in images. The science behind scene understanding involves pulling together the scattered efforts in computer vision, using methods from combinatorial optimization. This will have wide-ranging applications – for example, I have been working with a UK company Yotta on understanding road scenes for intelligent transport systems and with Sony on developing a general object or face tracker for use on their PlayStation 3 platform, which should be available in the next couple of years.
Read more about Professor Philip Torr's work at Oxford Brookes University.