Web 3D Virtual Fitting Room

I’m building a browser-based virtual fitting room where visitors receive precise size recommendations by dressing a fully adjustable 3D avatar that mirrors their own body. Users will upload a single full-body photo; the system must extract key measurements automatically, generate the avatar on-the-fly, and feed those dimensions into the fit algorithm.

Only body measurements need to be customizable for now—no facial or aesthetic tweaks—so the flow can stay fast and focused on accuracy. I expect the core to be a Three.js or Babylon.js scene (open to your preferred WebGL framework) tied to a lightweight machine-learning service for photo analysis. A Python/Node API that returns parsed measurements in real time would suit my stack, but let me know if you have a cleaner approach.

Deliverables I need to review before sign-off:
• Responsive web app with secure photo upload and deletion
• Automatic measurement extraction service with ≥90 % accuracy on standard test set
• Real-time 3D avatar generator matching those measurements (FBX, glTF or similar)
• Size-recommendation logic that maps avatar data to my existing product size chart (CSV provided)
• Brief deployment guide and commented source code

The experience should run smoothly on desktop and mobile browsers with no plug-ins. If you have prior work in virtual try-on, body-scan AI, or WebGL avatar creation, please include a link so I can see it in action.