Skip to main content

Frequently Asked Questions

Common questions about Visage.

General​

What is Visage?​

Visage is a local, privacy-first face recognition system designed as a prototype for assistive technology. It helps users identify people in real-time using computer vision.

Is Visage a medical device?​

No. Visage is an exploratory technical prototype, not a medical device. It is not FDA-approved and should not be used as a substitute for professional medical advice.

Is my data sent to the cloud?​

No. All data processing happens locally on your device. No data is sent to external servers.

Technical​

Which embedding model should I use?​

  • Facenet: Best default choice (fast and accurate)
  • VGG-Face: Use when accuracy is critical and speed is less important
  • ArcFace: Good balance of accuracy and speed
  • OpenFace: Use on resource-constrained devices

Why is the first identification slow?​

The first inference loads the embedding model into memory, which takes a few seconds. Subsequent identifications are much faster (sub-second).

Can I use Visage without a webcam?​

No, Visage requires a webcam for real-time identification. However, you can use the REST API to identify faces from static images.

Does Visage work with multiple faces?​

The system can detect multiple faces, but identifies them one at a time. Performance may degrade with many faces in frame.

Privacy & Security​

Where is my data stored?​

All data is stored in a local SQLite database at data/visage.db.

Is the database encrypted?​

Not by default. We recommend using full-disk encryption on your device.

Can I delete someone's data?​

Yes, you can delete any person's data through the UI or API.

Should I use Visage in public?​

No. Visage is designed for personal use with explicit consent from registered individuals.

Troubleshooting​

Backend won't start​

Ensure you've:

  1. Activated the virtual environment
  2. Installed all dependencies with pip install -r requirements.txt
  3. Initialized the database with python scripts/setup_db.py

Webcam not detected​

  • Grant webcam permissions in your browser
  • Ensure no other application is using the webcam
  • Try using localhost or HTTPS (required by some browsers)

No matches found​

  • Ensure good lighting conditions
  • Register multiple photos of each person
  • Adjust the MATCH_THRESHOLD in configuration
  • Verify embeddings are stored in the database

Slow performance​

  • Use a lighter model like Facenet instead of VGG-Face
  • Reduce frame rate in WebSocket client
  • Close other resource-intensive applications
  • Consider using GPU acceleration

Development​

Can I contribute to Visage?​

Yes! See the contributing guidelines in the repository.

How do I add a new embedding model?​

Visage uses DeepFace, which supports multiple models. You can switch models via configuration without code changes.

Can I deploy Visage to production?​

Visage is a prototype. For production:

  • Add authentication
  • Implement rate limiting
  • Use HTTPS
  • Add comprehensive logging
  • Consider legal/ethical implications

Ethics​

Anyone whose face you register in the system. Explain how Visage works and get explicit permission.

Can I use Visage for surveillance?​

No. Visage is designed for personal assistive use with explicit consent, not surveillance.

What about bias in face recognition?​

Face recognition systems can exhibit bias. We recommend:

  • Testing thoroughly with diverse datasets
  • Being aware of limitations
  • Not using for high-stakes decisions

Next Steps​