When it comes to sport, there is always a never ending debate as to who is the best. Now, it looks like these arguments can finally be put be bed thanks to a new AI system developed by researchers from Stanford University.
By using a database of annotated footage taken from broadcast video of tennis matches, the team of researchers were able to train their AI system to analyse the cyclical nature of tennis players. This allowed them to create a statistical model that predicts how tennis stars would fair if they face up against each other.
According to the research paper in which their finding were published “we use points from the video collection to model a player’s court positioning and shot selection decisions during points. We use these behavioural models to select video clips that reflect actions the real-life player is likely to take in a given match play situation, yielding sprites that behave realistically at the macro level of full points, not just individual tennis motions.”
By using this system, the researchers are able to a create a wide range of ‘what-if scenarios’ between players that would otherwise never meet each other on the court, according to Engadget.
By using previous footage of players like Novak Djokovic and Serena Williams for example, the system would be able to create a virtual match between the two super stars that would accurately represent how the two players players would react to different situations during a match.
The system isn’t perfect just yet though. During certain match ups you can see visual artefacts like changing lighting and player clothing suddenly change. The fact that the both the officials and audience don’t move during a match can also be quite distracting at time. Despite these issues the fact the system actually works is quite impressive, and could one day be used to finally settle the debate as to who the best really is.
Take a look at the AI system in action below:
Image credit: Pixabay