INTERNATIONAL
TEXT: MICHAELA NESVAROVA | PHOTO: GIJS VAN OUWERKERK >
AUTOMATED TRACKING OF
FOOTBALL
MATCHES There is a lot of money and interest in football. It is therefore highly important for football teams to pick the most suitable players. Now they have to do that based on a pure observation, but UT researchers and SciSports are working on developing a fully automated software for tracking football players and the ball. So far their efforts are paying off.
‘People love to bet on players’
T
he end goal is to do tracking in real time and to enable an automatic evaluation of players. You will be able to precisely tell what are their strengths and weaknesses, in which position they should play and so on,' explains Tauseef Ali, a postdoctoral researcher and the main coordinator of project BallJames, which aims to provide the best available tracking of football matches.
New tracking algorithm BallJames is a project funded by SciSports, a UT based company with the motto 'science serving football'. Most recently, the project has successfully reached the end of its first phase, during which the researchers managed to track players and the ball on a small playing field. 'We have an indoor football field with the size 10x20 meters. We use five cameras to record videos of the matches and analyze them afterwards,' describes Ali. 'Once we have the videos, we put them in the computer and get statistics regarding speed, hitting power etc. The process is fully automated. We use our own, completely new tracking algorithm to track position of the players' feet and the ball. We have manually placed a grid onto the playing field, so we can find the exact coordinates.' The research and technical
team is advised by Luuk Spreeuwers and Raymond Veldhuis from the University of Twente.
The next step Because the first part of the project has been a success also among its investors, who include international and national football players, BallJames is ready for the next step. 'The project has been extended and our goal now is to scale it up and be able to track matches on a real size football field and in real time,' says Ali. 'There are still a few challenges ahead of us. We need to be able to extract the vertical coordinate in case the players jump into the air, the ball bounces etc. This can be done using 3D tracking, which one of our researchers is working on.' If the team behind BallJames manages to overcome all the challenges, they can expect a great deal of interest from football clubs and sports enthusiasts. 'People who really like football are willing to spend a lot of money on it,' thinks Ali. 'The software could be used for live statistics during TV broadcasts. It also relates to gambling, people love to bet on players. At the moment, there are some products focused on tracking matches, but none of them are fully automated. This one is.' |
UT NIEUWS 10|2015 37