How it works
No sensors, no wiring, no second camera. Here's what happens between pressing record and getting the call.
Landscape orientation, both fencers and the scoring box in frame, ordinary gym lighting. Prop it on a bag or a bleacher — a few seconds around the touch is all Verdict needs.
Verdict locates 16 points on each fencer — body joints plus the blade's guard, midpoint, and tip — in every frame, and reads the scoring lights straight from the video.
You get the verdict, the action that produced it, and a skeleton-overlay replay showing exactly what the AI saw.
In sabre, when both scoring lights fire, the point doesn't go to whoever landed first by the clock — it goes to whoever had priority: the fencer who correctly started the attack, or who took it over with a parry or a well-timed counter. Reading priority means reading the whole phrase — who extended first, whether the attack fell short, whether a parry caught the blade — in a fraction of a second.
That's the judgment call referees train years to make, and the one that starts arguments at every club night when there's no referee on the strip.
Verdict approaches the problem the way a referee does, in three layers:
Every two-light call comes labeled with the action behind it, in the language referees actually use:
Because the call comes with its reasoning, every touch becomes a lesson — you see not just that you lost the touch, but why.
That's the whole setup. No sensors on the fencers, no wiring into the scoring box, no calibration step.
Early access opens this season. Free tier included.
Free tier at launch · One email when your spot opens — no spam.