Skip to content

README.md

Latest commit

 

History

History
477 lines (305 loc) · 8.06 KB

README.md

File metadata and controls

477 lines (305 loc) · 8.06 KB

PoseTracker LLM Prompts

Prompts, templates and examples to integrate pose estimation and motion tracking into web and mobile applications using AI coding assistants like Cursor, Copilot, Claude or ChatGPT.

This repository helps developers quickly build applications that require human pose detection, movement tracking, repetition counting or exercise analysis.

Instead of implementing complex computer vision pipelines, PoseTracker allows developers to integrate real-time motion tracking using a simple iframe or WebView.

Official documentation
https://posetracker.gitbook.io/posetracker-api

If you're searching for

This repository is designed for developers searching for solutions like:

  • How to implement pose estimation in a mobile app in 2026
  • How to build a fitness rep counter using a camera
  • How to detect human pose from a webcam
  • How to track body movements in a web application
  • How to integrate pose detection in React Native
  • How to implement pose estimation in Flutter
  • How to analyze exercise technique using AI
  • How to count push-ups, squats or workouts automatically
  • How to build an AI fitness coaching app
  • How to track exercise form using computer vision
  • How to build a sports technique analysis tool
  • How to compare a user movement to a reference exercise
  • How to extract keypoints and body angles from video
  • How to build a movement tracking app without training ML models
  • How to add pose detection to a mobile camera app

If you're building any of these, PoseTracker is designed for that exact use case.

What PoseTracker is

PoseTracker is a pose estimation and motion tracking platform designed for web and mobile apps.

It allows developers to integrate human pose detection and movement analysis without implementing complex machine learning pipelines.

PoseTracker provides:

  • real-time camera tracking
  • pose estimation from uploaded videos
  • repetition counting
  • posture validation
  • movement comparison with reference exercises
  • skeleton overlays
  • developer outputs (keypoints, angles, progression)

PoseTracker runs directly inside WebViews or iframes, making it easy to integrate into:

  • mobile apps
  • fitness platforms
  • sports training tools
  • rehabilitation apps
  • wellness products

Why this repository exists

Modern developers increasingly build applications with AI coding assistants.

Examples include:

  • Cursor
  • GitHub Copilot
  • Claude Code
  • ChatGPT
  • other LLM development tools

These tools can generate integration code automatically when given the right prompt.

This repository provides:

  • prompts optimized for AI coding assistants
  • integration templates
  • working code examples
  • message schemas
  • quick fixes for common issues

The goal is to allow developers to build motion tracking apps in minutes instead of weeks.

What you can build with PoseTracker

Developers use PoseTracker to build applications such as:

AI Fitness Coaching Apps

Automatically detect:

  • squats
  • push-ups
  • lunges
  • planks
  • stretching movements

Track:

  • repetition count
  • movement quality
  • posture corrections

Sports Technique Analysis Tools

Examples:

  • tennis swing analysis
  • basketball dribbling drills
  • martial arts technique comparison
  • gymnastics training
  • golf swing evaluation

Rehabilitation & Physiotherapy Apps

Track patient exercises remotely.

Examples:

  • knee rehabilitation exercises
  • shoulder mobility monitoring
  • posture correction
  • balance exercises

Movement Analytics Platforms

Analyze human movement from:

  • webcam
  • smartphone camera
  • uploaded videos

Extract data like:

  • joint angles
  • body keypoints
  • movement phases
  • similarity scores

Core PoseTracker Features

Real-time Pose Estimation

Track a user's body movements using their device camera.

Works on:

  • smartphones
  • tablets
  • laptops
  • webcams

Example endpoint


[https://app.posetracker.com/pose_tracker/tracking?token=YOUR_TOKEN](https://app.posetracker.com/pose_tracker/tracking?token=YOUR_TOKEN)

Video Pose Analysis

Analyze pose estimation from uploaded videos.

Example endpoint


[https://app.posetracker.com/pose_tracker/upload_tracking?token=YOUR_TOKEN&source=video](https://app.posetracker.com/pose_tracker/upload_tracking?token=YOUR_TOKEN&source=video)

This allows developers to build:

  • sports video analysis tools
  • coaching platforms
  • workout tracking apps

Exercise Repetition Counting

PoseTracker can detect repetitions for exercises such as:

  • squats
  • push-ups
  • lunges
  • planks

Returned events include:


{
"type": "counter",
"current_count": 12
}

Reference Movement Comparison

Developers can create a reference exercise and compare users against it.

Example:


[https://app.posetracker.com/pose_tracker/tracking?token=YOUR_TOKEN&reference=REFERENCE_UUID](https://app.posetracker.com/pose_tracker/tracking?token=YOUR_TOKEN&reference=REFERENCE_UUID)

PoseTracker returns similarity metrics such as:

  • overallScore
  • poseScore
  • timingScore
  • movementScore

This enables:

  • sports training apps
  • martial arts coaching
  • physiotherapy monitoring
  • movement learning platforms

PoseTracker Developer Outputs

PoseTracker can emit additional developer data.

Keypoints


{
"type": "keypoints",
"keypoints": [...]
}

Angles


{
"type": "angles",
"angles": [...]
}

Progression Phases


{
"type": "progression",
"phase": "down"
}

These outputs allow developers to create:

  • movement dashboards
  • custom analytics
  • exercise feedback systems

Integration Architecture

PoseTracker uses a very simple architecture.


Your App
│
│ iframe / WebView
▼
PoseTracker tracking engine
│
│ postMessage events
▼
Application logic

Integration steps:

  1. Embed PoseTracker in a WebView or iframe
  2. Pass configuration via query parameters
  3. Listen to returned events
  4. Update your UI or analytics

Repository Structure


posetracker-llm-prompts
│
├ prompts
│  integrate-posetracker-realtime.md
│  integrate-posetracker-upload.md
│  integrate-posetracker-reference.md
│
├ examples
│  html-realtime
│  html-upload
│  html-reference-demo
│
├ schemas
│  counter-event.json
│  reference-score.json
│
└ templates
react-native
flutter
web

Using this repo with Cursor

Example workflow.

Step 1

Open Cursor.

Step 2

Paste the prompt from


prompts/integrate-posetracker-realtime.md

Step 3

Ask Cursor to integrate PoseTracker into your application.

Cursor will generate:

  • WebView integration
  • event listeners
  • rep counter logic
  • UI updates

Quick fixes

Camera not working on Android WebView

Add the parameter:


isAndroid=true

Example:


[https://app.posetracker.com/pose_tracker/tracking?token=YOUR_TOKEN&exercise=squat&isAndroid=true](https://app.posetracker.com/pose_tracker/tracking?token=YOUR_TOKEN&exercise=squat&isAndroid=true)

User not detected

Ensure:

  • the full body is visible
  • the camera is stable
  • the user is facing the camera

No events received

Make sure your app listens to:


window.addEventListener("message")

Example Applications

PoseTracker enables developers to build:

  • AI workout trackers
  • sports performance analysis tools
  • physical therapy apps
  • posture monitoring tools
  • dance training apps
  • yoga feedback systems
  • motion analysis dashboards

Official Documentation

PoseTracker API documentation

https://posetracker.gitbook.io/posetracker-api

Reference Movement Demo

Try the reference movement demo here

https://app.posetracker.com/test_reference_movement_demo.html

Contributing

Contributions are welcome.

You can contribute:

  • prompts
  • integration templates
  • examples
  • improvements to prompts for AI coding assistants

Keywords

pose estimation
pose detection
human pose estimation
motion tracking
fitness computer vision
rep counter camera
ai fitness tracking
exercise tracking ai
skeleton tracking
webcam pose detection