VISPATH
Vision-based Internet Software Platform for the Applied Tracking of Humans
Home | Demos | Publications | Press | Contact

Press

This section of the website contains press output related to the project, including press releases and newspaper cuttings.

28 February 2003: Dundee University Press Office

Computers mimicking movement

Forget kids sitting motionless in front of their computers with a console controlling the action. When a researcher at the University of Dundee proves his new idea works, they will be able to join in the on-screen action.

Stephen McKenna and his team have just secured money from the Scottish Executive to prove their concept for software to mimic human movements in the Proof of Concept funding round. He says that an image on a computer screen will be able to show the movement of people in front of it - running, jumping, waving - any motion - using just a domestic PC and camera - without any special clothing or sensors on their bodies.

And more than mimicking movement, the beauty of this new software is that it does not merely track and mimic the visual, but collects and analyses data from the movement it sees. Imagine a golfer practising his swing. As well as being able to watch himself preparing and following through, the software will produce an accurate analysis of the all important physics of his movement so he can compare the essentials of his swing with the pros. Stephen and his research associates Gordon McAllister and Tim Roberts think that their software will be useful not only in the entertainment industry, but could be used in the health sphere to screen people's walking for problems. Through this gait analysis, doctors could determine the nature of a walking defect.

This is just one of the three projects at the University of Dundee awarded funding in the Proof of Concept round. Two other projects in the Faculty of Engineering and the Faculty of Life Sciences have been awarded funding to prove their technology.

By Jenny Marra, Press Officer 01382 344910 j.m.marra@dundee.ac.uk

28 February 2003: Scottish Enterprise

Potential of technology: Forget kids sitting motionless in front of their computers with a console controlling the action. When this new technology proves its concept, they will be able to join in the action. The image on a computer screen will mimic movement of people in front of it - running, jumping, waving - any motion - all with just a domestic PC and camera and without any special clothing or sensors on their bodies.

And more than mimicking movement, the beauty of this new software is that it does not merely track and mimic the visual, but collects and analyses data from the movement it sees. Imagine a golfer practising his swing. As well as being able to watch himself preparing and following through, the software will produce an accurate analysis of the all important physics of his movement so he can compare the essentials of his swing with the pros.

Not just for the entertainment industry, this software could be used in the health sphere to screen people's walk for problems. Through this gait analysis, doctors could determine the nature of a walking defect.

Summary of technology: Hardware - domestic camera and PC.

State-of-the-art algorithms that share a common mathematical framework based on probability and models of articulated body structure and learned visual appearance.

Concept to be proven: The concept we seek to prove is that this software based on mathematical principles will be a flexible product able to support an exciting range of commercial applications. The software will be able to view a moving person and not only record but interpret his or her movement, identifying the movement, assessing its implications and reacting appropriately, possibly creating an image on screen of a character mimicking the movement or reacting to it.

The software should be flexible and general enough to be used for a number of purposes including entertainment, medicine and sport.

A full-body tracker automatically estimates the 3D position and orientation of the torso as well as the 3D joint angles of the shoulders, elbows, hips and knees.

Identifying a pointing gesture.