Scene Understanding using New Global Energy Models-a project on machine learning/computer vision

This project is about computer vision and machine learning.

The computer vision group at Oxford Brookes has won multiple international awards for its research work

The Computer Vision group in the Department of Computing was formed in 2005 by Philip Torr and William Clocksin, and is led by Philip Torr. It comprises 8 PhD's (with vacancies for two more if you know any bright applicants) and 4 post docs, in addition we have some semi regular visitors and joint grants with other universities that account for another 3 PhD's and 2 postdocs.

The aim of the group is to engage in state of the art research into the mathematical theory of computer vision and artificial intelligence, but to keep the mathematical research relevant to the needs of society. Our research is focused on Bayesian methods, in particular the study of the mathematics underlying Markov Random Fields, combinatorial optimization and Bayesian nets.

The applications come in many forms, and we are involved with several major companies and organizations. With Sony we are working on human computer interaction (via a camera, the "EyeToy") for the Play Stations 2 and 3, with Sharp we are working on generation of content for 3D displays, with Oxford Metrics Group we are working on computer understanding of films (e.g. what is the shape of objects in the scene etc) in order to make better special effects, we also work on motion capture of humans (and animals) in order to drive computer generated avatars. We work on medical image analysis and on surveillence. We also do collaborative work with Microsoft Research, London, Cambridge and Oxford Universities
.

See here http://cms.brookes.ac.uk/staff/PhilipTorr/



This proposal concerns scene understanding from video. Computer vision algorithms for individual tasks such as object
recognition, detection and segmentation has now reached some level of maturity. The next challenge is to integrate all these
algorithms and address the problem of scene understanding. The problem of scene understanding involves explaining the
whole image by recognizing all the objects of interest within an image and their spatial extent or shape in 3D.
The first application to drive the research will be the problem of automated understanding of cities from video
using computer vision, inspired by the availability of massive new data sets such as that of Google’s Street View
http://maps.google.com/help/maps/streetview/, Yotta http://www.yotta.tv/index.php (who have agreed to supply Oxford
Brookes with data) and Microsoft’s Photosynth http://labs.live.com/photosynth/. The scenario is as follows: a van drives
around the roads of the UK, in the van are GPS equipment and multiple calibrated cameras, synchronized to capture and
store an image every two metres; giving a massive data set. The task is to recognize objects of interest in the video, from
road signs and other street furniture, to particular buildings, to allow them to be located exactly on maps of the environment.
A second scenario would be to perform scene understanding for indoor scenes such as home or office, with video taken from
a normal camera and Z-cam.