A computer vision application that calculates body measurements from a single photo using OpenCV and MediaPipe pose detection.

• Upload a person's photo
• Automatic pose detection using MediaPipe
• Calculate key body measurements:
  • Height
  • Shoulder width
  • Chest/Bust circumference
  • Waist circumference
  • Hip circumference
  • Arm length
  • Leg length (inseam)
  • Neck circumference
  • Wrist circumference
  • And more tailoring measurements

1. Pose Detection: Uses MediaPipe to detect 33 key body landmarks
2. Reference Scaling: Uses known body proportions or user-provided reference measurements
3. Anthropometric Calculations: Applies standard body measurement formulas
4. Real-time Processing: Processes images and returns measurements instantly

1. Upload a clear, full-body photo (front-facing preferred)
2. Optionally provide a reference measurement (like height) for better accuracy
3. Get comprehensive body measurements suitable for tailoring

• Backend: Hono.js with pose detection and measurement calculations
• Frontend: React with image upload and results display
• Computer Vision: Currently uses mock pose detection (see integration guide below)
• Calculations: Anthropometric formulas for body measurements
• Styling: TailwindCSS for responsive design

├── backend/
│   ├── index.ts              # Main API server
│   ├── pose-detection.ts     # MediaPipe pose detection
│   ├── measurements.ts       # Body measurement calculations
│   └── utils.ts             # Helper functions
├── frontend/
│   ├── index.html           # Main page
│   ├── App.tsx              # React app component
│   └── components/          # UI components
└── shared/
    └── types.ts             # Shared TypeScript types

• POST /api/measure - Upload image and get body measurements
• POST /api/report - Download measurements as text file
• GET /api/health - Health check endpoint
• GET /api/docs - API documentation
• GET / - Serve the main application

This demo currently uses mock pose detection for demonstration purposes. To integrate with real pose detection:

npm install @mediapipe/pose @mediapipe/camera_utils

npm install @tensorflow/tfjs-node @tensorflow-models/posenet

npm install opencv4nodejs

See /backend/real-pose-integration.ts for detailed implementation examples.

The system currently:

• ✅ Provides a complete web interface for image upload
• ✅ Validates image format and size
• ✅ Implements comprehensive measurement calculations
• ✅ Uses anthropometric formulas for accurate estimations
• ✅ Provides confidence scoring
• ✅ Generates downloadable measurement reports
• ⚠️ Uses mock pose detection (needs real implementation)

The measurement calculations are based on established anthropometric research and provide:

• Primary measurements: Height, shoulder width, chest, waist, hip circumferences
• Arm measurements: Full arm, upper arm, forearm lengths and circumferences
• Leg measurements: Leg length, inseam, thigh, calf circumferences
• Tailoring measurements: Neck, armhole, back/front widths, and key distances

Accuracy depends on:

1. Photo quality: Clear, well-lit, full-body images
2. Pose quality: Upright stance with arms slightly away from body
3. Reference measurement: Providing actual height significantly improves accuracy
4. Pose detection quality: Real implementation will be more accurate than mock