Live Demo: https://kegangwangccnu.github.io/FacePhys-Demo/

Note: A proxy might be required in restricted network regions.

FacePhys Demo is a browser-based remote photoplethysmography (rPPG) monitor. It uses State Space Models (SSMs) to extract heart rate and heart rate variability from facial video feeds via a standard webcam.

The application runs inference locally using LiteRT and WebAssembly. It is designed for edge environments, offering low latency (~5ms) with a model size of approximately 4MB.

• Zero Data Upload: All inference occurs on-device.
• No Cloud Processing: Video data and physiological signals are never transmitted to a server.
• Offline Capable: The application functions without an internet connection once loaded.

• State Space Models: Models heart dynamics as continuous-time controlled ODEs to handle irregular sampling.
• Multi-threaded Architecture: Uses Web Workers for inference, PSD calculation, and plotting to maintain UI responsiveness.
• Visualization: Displays real-time attention heatmaps, 3D state trajectories, and BVP/PSD waveforms.
• Interface: Windows 95/98 style UI.

This project is a Progressive Web App (PWA) and can be installed as a standalone application.

1. Open the website.
2. Click the Install icon in the address bar.

Scan the QR code below to access the application on your mobile device:

1. Open the scanned link in Chrome.
2. Open the menu and select Add to Home screen or Install App.

1. Open the scanned link in Safari.
2. Tap the Share button and select Add to Home Screen.

MIT License with Privacy Protection Addendum — see LICENSE for details.