Scoring goals

Computer scientists at Kingston University are developing computer models that can assess the effectiveness of sporting tactics.

With help from funding of nearly £83,000 from the EPSRC, computer scientists at Kingston University are developing models that can assess the effectiveness of sporting tactics.

The project will use digitised conventional video footage as the basis for the new computer models. A purpose-built visual tracking software will then identify the players, record their positions and track them throughout play, and their positions will be fed into predictive models trained to recognise known tactics and strategies.

The models will aim to recognise the difference between attack and defence and the dominance of one team over another, and will allow the user to explore the effect of different tactics and of different strategies to counter them.

As well as offering the potential for much more sophisticated analysis of players than can be achieved observing video footage alone, computer models are also much quicker to use. They avoid the need to edit and hunt through huge volumes of tape, and include search software that can rapidly scan for examples of specific tactics or patterns of play.

In the past, developing this kind of computer model has been difficult because sporting tactics involve a lot of co-ordination between players. The project will tackle this by harnessing models specially designed to handle situations where the movements of two or more co-ordinated objects need to be tracked at the same time.

The research will start by focusing on tennis singles and doubles and then look at more complex sports such as football and basketball.

Dr. Ahmed Shihab of the School of Computing & Information Systems is leading the project.

“As well as helping specialised sports training, the technology we are developing could have benefits in fields such as realistic computer gaming, virtual reality and surveillance, which also involve co-ordinated human activity,” Shihab said.

Once developed, the researchers may deploy the system in a commercial role to automatically generate highlights from tennis footage.

LampLighters, a commercial partner with a broadcasting industry background, aims to evaluate the application.