Wheelchair Football Association expands AI initiative into live streaming By David Davies Tuesday, August 5, 2025 - 09:10
Print This Story
At the SVG Europe Football Summit earlier this year, Alex Dowding, vice secretary of the Wheelchair Football Association (WFA), highlighted the organisation's adoption of new technologies in order to streamline production and make the workflow more accessible for its volunteer-run operation. Now the WFA has revealed the latest developments in its long-term WFAi initiative, which is designed to revolutionise how we share, broadcast, view and analyse powerchair football , according to Jack Humphries, the WFA's national league secretary.
In a recent LinkedIn post, Humphries said that the organisation was proud to be at the forefront of using cutting-edge technology and AI to enhance the beautiful game we love . Noting that the WFA has already begun implementing WFAi in its Score Line app, which provides intelligent match reports, he said that the next step in the journey entailed expanding the capabilities of our live streaming systems with AI .
Speaking to SVG Europe, Humphries and Dowding expanded upon the key objectives of WFAi. About two years ago we were starting to have the foundations for using technology to build a better data set, and over time this has [dovetailed with WFAi] as improving the data we collect is one of the main goals for this initiative, says Humphries. We're a powerchair football organisation and we're always trying to think about how we can collect more data for coaching analysis. But we're also always looking at engagement and the viewing experience in the app and online streams, and in that regard we've got a lot of data from tracking players and tracking the ball [that we can utilise].
A current area of focus is on implementing those learnings in live stream workflows, supporting a single camera - and, in time, multiple cameras - that can track the game. At which point you've got a semi-automated, or fully automated, live streaming workflow that's powered by AI, notes Humphries.
Origins of WFAi
While the WFAi is the latest phase of the organisation's technological trajectory, Dowding indicates that it is a continuation of an ethos established in the earliest days of the WFA. We made a decision then that we would only implement changes to our technology and workflows if they improved production, live streaming, and the ease with which we could do administration. You could make an improvement as long as it didn't affect accessibility, and you could improve accessibility as long as it didn't affect performance or production, although ideally you were doing things that improved accessibility and production. [Ultimately] that pushed us towards a scenario where we had our own fast single point of data containing everything from letters to go to people if they've got a yellow card, through to information about the number of appearances people have made - and it's that resulting ability to capture data quickly from multiple directions that's provided the opportunity to develop WFAi.
We've got so many different angles through which we can improve the game by using new technologies. It's shaping up to a very exciting period of development for the WFA
One consequence of this strategy has been an expanded family of apps, including WFA Touchline, where referees and match officials can log their data, and Match Day Live, which allows an entire stream to be prepped and initiated from an iPad. We preload all the stream keys and pre-schedule everything on YouTube, so [on the day] it's possible to select the game from drop-downs, enter name details, set the match data, and then hit start streaming' - and the game will start up and go to the right endpoint on YouTube. During the game it means you get normal football graphics with goal scorers, yellow cards, half-times, a clock and so on, adds Dowding.
This can already be achieved in conjunction with the use of a static camera, but with WFAi enabling ball-tracking there is an expectation of automated production switching between multiple cameras in the not-too-distant future. The plan is for three cameras on a pitch, says Humphries, so the main camera doing the panning left and right, and then two individual PTZ cameras elsewhere. If we know the lineups at the start of the game we can track individual players, and so we can consider a production workflow where the commentator can choose a player for a player cam and select them by name. The camera will know where they are and be able to track them using the ball tracking algorithm and player tracking information.
It is hoped that, in time, this level of automation will make it easier to achieve two or even one-person production per game. But there is still room for refinement, confirms Humphries: Over the summer I think we're going to look at tuning the ball-tracking a bit more. For instance, we've been finding that occlusion is quite a big problem; when the ball goes behind a chair, you're having to imagine where the ball is for a moment, and then it pops out somewhere else. It could be that you have a situation where the ball goes the entire length of the pitch, occluded, and you're going where is the ball?' But we've got a few ideas for resolving that situation and [in the next few months] I think we should know which solutions work the best.
Match reports
Another area of interest for the WFAi project is the automatic creation of match reports, using discrete information from the database and event commentary transcription. You then ask AI to prioritise the discrete data in terms of facts, but within the narrative taken from the transcript, says Dowding. You obviously prioritise having players' names sp
