Product · Live · AI & Computer Vision

Posemetric - AI-Powered Equine Gait and Movement Analysis

Film a horse trotting with a phone, and Posemetric turns the clip into frame-by-frame joint angles, stance timing and an asymmetry report - the same data a lameness exam looks for, without a motion-capture lab.

Posemetric analysis frame showing a trotting horse with tracked carpus and tarsus joints, live joint-angle readouts and per-limb angle charts
Project Overview

The essentials, at a glance.

IndustryEquine Health & Agri-Tech
SolutionWeb App + Android App
RoleComputer Vision, Full-Stack Development
StatusLive at app.posemetric.com
The Challenge

Gait asymmetry is hard to track between vet visits.

Subtle lameness and gait asymmetry in horses are usually judged by eye, either by a vet during a scheduled visit or by an experienced owner who happens to notice a change. Objective motion-capture equipment exists, but it lives in clinics and research labs - not in a stable yard. Owners and trainers had no simple way to capture the same kind of joint-angle data themselves, on their own schedule, with a device they already carry.

Our Solution

A phone video in, a kinematic report out.

Posemetric EquiGait takes a short clip of a horse trotting - filmed side-on, following the in-app filming guide - and runs it through a computer vision pipeline that tracks anatomical landmarks frame by frame. The result is a set of joint-angle curves and a stance-by-stance breakdown the owner can read directly, or export and send to their vet. The same account and backend work identically on the web app and the companion Android app.

Key Features

What the platform does.

Guided Video Capture

An in-app filming guide shows good, too-far and off-angle examples so uploaded clips are actually analyzable before they're sent.

Automatic Landmark Tracking

The pipeline locates and tracks the horse's limb joints - carpus and tarsus on both sides - through every frame of the clip.

Frame-by-Frame Joint Angles

Angle curves for LF/RF carpus and LH/RH tarsus, plus stance detection (e.g. "RF+LH"), are plotted across the full clip.

Asymmetry Reporting

A dedicated report compares left and right sides to surface the kind of asymmetry a lameness exam looks for.

Studio Dashboard

Upload a clip, watch it process, and come back to browse every past analysis for a horse in one place.

One Account, Two Apps

Firebase Authentication is shared between the web app and the native Android app, so a single sign-in covers both.

Technical Approach

How it's built.

A React web client and a native Android app both talk to the same Cloudflare Worker API, so a clip uploaded from either one lands in the same account and the same analysis history. The web app is a static SPA, deployed automatically on every push to main.

React 19 Vite TypeScript Tailwind CSS 4 Firebase Authentication Cloudflare Workers API Cloudflare Pages GitHub Actions CI/CD Android (companion app)
Results

Where the project stands.

Live in production

Running today at app.posemetric.com, with continuous deployment from the main branch on every release.

Two clients, one backend

The web app and the Android app share one Cloudflare Worker API and one Firebase account, so analysis history stays in sync across devices.

Real pipeline output

Every screenshot on this page is genuine output from the analysis pipeline, not a mockup - people in the frame are blurred for privacy.

Filming Guide

What the pipeline needs from a clip.

Analysis quality depends on how the clip is filmed. The app guides owners toward a clean side-on trot pass and away from common mistakes.

Example of a well-filmed clip: full horse body visible, side-on, at trotting distance

Good: side-on, full body

The whole horse stays in frame through the trot pass, filmed roughly perpendicular to the direction of travel.

Example of a poorly-angled clip filmed from an oblique angle rather than side-on

Avoid: filmed at an angle

An oblique camera angle distorts the joint geometry the pipeline is measuring.

Example of a clip filmed from too far away, making the horse too small in frame

Avoid: too far away

If the horse is too small in frame, landmark tracking loses the precision it needs.

Need a similar AI or web-based solution?

Tell us briefly about your project and timeline. You'll receive an initial technical assessment and a proposal for next steps.

Discuss Your Project