Now in early access
Know every jump, without watching every second.
Levijoy turns hours of training and competition footage into an accurate jump count and timeline in minutes — so your yard spends less time scrubbing through video and more time in the saddle.
- The problem
Counting jumps by eye means watching everything twice.
After every session, someone has to scrub back through the footage to work out how many fences were jumped, when, and how it went — time that could be spent coaching, riding, or reviewing what actually matters.
- The Levijoy way
Point it at the footage. Get the jump list back.
Levijoy tracks the horse through the video and works out exactly when each jump happened, so your team gets a clean timeline instead of a stopwatch and a notepad.
Automatic jump counting
Get an accurate jump count and timestamp for every fence, without touching the timeline yourself.
Works with your existing footage
Built on pose-tracking, not special cameras — it works with the training and competition video you already record.
Handles doubles and combinations
Recognises closely-spaced efforts in combinations, not just single fences on their own.
Ignores canter strides
Filters out regular canter strides and camera movement, so you're not chasing false alarms.
How it works
From raw footage to a jump list, in four steps.
Track
The horse is tracked frame by frame through your video.
Clean
Camera movement and tracking glitches are filtered out automatically.
Detect
Candidate jumps are picked out from the motion pattern.
Confirm
Canter strides and false alarms are filtered, leaving the real jumps.
Try Levijoy on real competition footage.
Pick a sample round and watch Levijoy work through it — no upload, no waiting, just the pipeline in action.
Tested, not just trained
Evaluated on real rounds — including horses and riders it had never seen before.
- real competition rounds evaluated
- 15
- of jumps caught in tuned footage
- 89%
- evaluation tiers, including full hold-out riders
- 3
Why Levijoy exists
Built by a rider, for riders.
Levijoy grew out of a University of Warwick research project. To build and train it, I used competition and training footage from several different horses — including three of my own — plus footage of myself, my sister, and friends competing, turning it into a tool built for real yards, not just a dissertation.
Ready to bring AI jump analysis to your yard?
We're onboarding a small number of yards during early access — get in touch and we'll set up a demo with your own footage.