Client-side face detection with TensorFlow.js

Client-side face detection with TensorFlow.js with Blazeface model to detect faces via webcam feed or uploaded photo. No installation required - give it a go.

Features

1. Face Detection:
   Process video feed from your webcam with detect multiple faces in real-time
2. Image Upload:
   Analyze faces in uploaded photos, large images are tricky, working on a fix
3. Browser-based and privacy-focused:
   No installation required - works directly in your browser client side. All processing happens locally and no data leaves your device
4. Download Canvas:
   Download the canvas as an image. This is a screenshot of the current state of the canvas.

How It Works

This app uses TensorFlow.js to run the Blazeface neural network model directly in your browser. The model has been optimized for real-time face detection while maintaining accuracy. All processing happens locally on your device - no images or video are sent to any server.

Privacy Policy

This app processes all images and video locally in your browser. No data is sent to any server or stored anywhere. The webcam feed is only used for real-time detection and is never recorded or transmitted.

Stack

• Next.js 15 and TensorFlow.js 4.22.0
• Uses the Blazeface model for fast, accurate face detection

Will work on:

• Camera selector
• Explore other models now that blazeface is deprecated and replaced by this module.

Note: This is an experiment and may not work on your device.

If you experience any issues, please get in touch. Lemon picture shown.