What exactly is allez go used for?

When people watch a fencing bout for the first time, they often walk away with more questions than answers. “Why is the point system so complicated?” “Doesn’t the first person that hits get the point?” “What’s right of way?” Public interest in fencing quickly dissipates as viewers often cast fencing as an impossibly complicated, niche sport. Allez Go will be used in fencing broadcasts and live streams and will generate priority indicators in real-time. While this won’t magically make fencing a nationally popular sport, it will help introduce new viewers to the concept of right of way and how fencers play with priority. It will reveal a whole new dimension of fencing to spectators.

How does allez go work?

Using machine learning techniques, Allez Go was built to imitate how human referees decide priority - by using both vision and hearing. To handle the visual aspect, Allez Go uses pose estimation to get the precise position of each fencer’s body. Allez Go then uses a complex neural network to process visual and audio to generate a priority indicator.

is allez go responsible?

We are committed to ensuring that artificial intelligence is used responsibly and ethically in fencing. Allez Go will not be used as a replacement for human referees and our official policy is that any human referee will always have more authority than Allez Go. We do not support the use of AI referees to argue with human referees. We’ve also made several steps to ensure that Allez Go is not biased towards any particular club, nation, or a fencer. Our data is an uncensored collection of official FIE bouts and bouts were not excluded from the dataset due to gender, nationality, or any other factor. Our dataset is also available for 3rd-party review upon request.

Does Allez go directly affect the bout?

No, Allez Go is only for spectators watching from a Youtube Livestream or replay of the bout. It will not be visible to spectators that are in-person, at the venue and should have no influence on the human referee’s calls in the bout. Allez Go will not be used to award points to fencers or make any actual calls.

What are The hardware specifications?

For the best results, Allez Go requires a camera with a clear, unobstructed view of the strip mounted near the center of the strip (similar to ones already in use at most major tournaments). It also requires a microphone located near the strip with preferably little background noise. Computationally, Allez Go requires a GPU processor in order to run in real-time.

Does Allez Go work for Saber?

Unfortunately, Allez Go is trained only to work on foil bouts. However, bringing Allez Go to saber will be a top priority once Allez Go starts being used in tournaments.

What are allez go’s other potential use-cases?

Allez Go can make regular calls such as touch left or touch right but unfortunately, is unable to provide the exact reason for that call. For example, it may say an action was ‘touch right’ but won’t be able to say that it is a ‘parry riposte from right, touche’. In the long run, AI referees may be used to serve as backup referees or for video challenges but we believe the technology still has a long way to go before it can fulfill this role reliably.

Did we miss your question?