Capture

• Know cameras grab an image or video
• Assume capture happens without your consent
• Expect high-quality frames feeding the system

Detection

• The system locates a face in frame
• It pins key facial landmarks
• It isolates your face from the scene

Feature Extraction

• Unique features get measured precisely
• Measurements convert into a digital faceprint
• The template encodes your identity numerically

Matching

• Faceprint is compared against stored templates
• Each comparison returns a confidence score
• The system seeks the closest match

Decision

• A score above threshold confirms identity
• Matches above threshold grant access
• Low confidence rejects or flags you

The Database

• Encrypted templates sit ready for comparison
• Stored faceprints persist far beyond capture
• A security layer guards against spoofing

Where Deployed

• Watch for access control at buildings
• Expect device unlock on phones and laptops
• Assume surveillance scans for persons of interest

Privacy Stakes

• Demand consent and data minimization
• Question retention and bias in the model
• Treat every faceprint as a lasting risk