Pose Analysis Library

This library provides real-time human pose analysis using MediaPipe. It includes React components for video and live camera analysis, along with standalone logic for counting repetitions and providing feedback for specific exercises.

📂 Project Structure

libraries/pose-analysis/
├── src/
│   ├── analyzers/         # Core exercise logic (pure TypeScript)
│   │   ├── squat.ts       # Squat form analysis
│   │   ├── kettlebell.ts  # Single Hand Snatch analysis
│   │   ├── long_cycle.ts  # Long Cycle (Double Kettlebell) analysis
│   │   └── types.ts       # Shared interfaces
│   ├── components/        # React UI components
│   │   ├── AnalysisPage.tsx # Video/Image upload & analysis
│   │   ├── ExercisePage.tsx # Live webcam exercise tracking
│   │   └── BottomNav.tsx    # Navigation component
│   ├── hooks/             # Custom React hooks
│   │   └── usePoseDetection.ts # MediaPipe integration hook
│   ├── lib/               # Utilities
│   │   └── utils.ts       # Helper functions (cn, etc.)
│   └── index.ts           # Public API exports
├── dev-app/               # Standalone Vite development app
├── start-dev.sh           # Script to build & run dev-app
├── package.json           # Library dependencies
└── tsconfig.json          # TypeScript configuration

🏋️ implemented Exercises

1. Squat Form (squat.ts)

• Logic: Calculates the angle between Hip, Knee, and Ankle.
• States: UP -> DOWN -> UP.
• Feedback: Checks depth (< 90 degrees) and full extension.

2. Single Hand Kettlebell Snatch (kettlebell.ts)

• Logic: Tracks wrist position relative to shoulder (top) and hip (bottom).
• States: Independent UP/DOWN states for Left and Right arms.
• Feedback: "Left Up!", "Right Rep Complete!", etc.

3. Long Cycle (Double Kettlebell) (long_cycle.ts)

• Logic: Tracks BOTH hands simultaneously.
• Requirement: Both wrists must move from below hips to above shoulders and back down together.
• Feedback: "Lift both hands!", "Rep Complete!".

🚀 Standalone Development

You can run this library in isolation without the main app.

1. Navigate to directory:

   cd libraries/pose-analysis

2. Run the dev script:

   ./start-dev.sh
   # Or on Windows Git Bash:
   bash start-dev.sh

   This will:
   • Install dependencies for the dev-app.
   • Build the library.
   • Start a local Vite server at http://localhost:5173.

🔌 Integration Guide

Frontend (React App)

To use these components in the main baas application:

1. Add Dependency: Add to root package.json:

   "@lid-vizion/pose-analysis": "workspace:*"

2. Install:

   npm install

3. Import & Route: In src/App.tsx:

   import { AnalysisPage, ExercisePage } from "@lid-vizion/pose-analysis";
       
   // In Routes:
   <Route path="/analysis" element={<AnalysisPage />} />
   <Route path="/exercise" element={<ExercisePage />} />

Backend (AWS Lambda)

To run analysis logic on the server (e.g., for processing uploaded videos):

1. Handler: A handler file has been created at lambda/src/handlers/pose-analysis.ts.

   • It imports the analyzers (SquatAnalyzer, etc.) directly.
   • Uncomment the code in this file to enable it.

2. Infrastructure (SAM): A resource definition has been added to lambda/workspace-template.yml.

   • Uncomment the PoseAnalysisFunction section to provision the Lambda.

3. Dependencies: Add @lid-vizion/pose-analysis to lambda/package.json if you want to share the exact same logic code between frontend and backend.

🛠 Dependencies

• @mediapipe/pose: AI model for body tracking.
• react-webcam: For handling camera input.
• lucide-react: UI icons.
• tailwindcss: Styling (via clsx and tailwind-merge).