Glossary

Face recognition glossary.

Plain-English definitions of the terms you'll meet when building with biometrics — from embeddings to BIPA.

1:1 verification

Comparing two face images to decide whether they show the same person — for example a selfie against an ID photo.

Searching one face against a gallery of N enrolled people to find who it is, if anyone.

A liveness check that asks the user to do something — blink, turn or follow a dot.

A compact numeric representation of a face used for matching. It cannot be reversed into the original photo.

Illinois' Biometric Information Privacy Act — requires notice, written release and a public retention schedule.

A number, usually 0–1, expressing how strongly the system believes two faces match.

Data Protection Impact Assessment — required under GDPR for high-risk processing such as large-scale biometrics.

India's Digital Personal Data Protection Act, 2023 — a consent-centred data protection law.

Registering a person by creating and storing their face template for later matching.

Finding where faces are in an image, returned as bounding boxes and landmarks.

The vector a neural network produces for a face; similar faces have nearby embeddings.

How often the system wrongly declares two different people a match. Also called FAR.

How often the system wrongly rejects two images of the same person. Also called FRR.

Feeding synthetic or deepfake video directly into the capture pipeline, bypassing the camera.

The international standard for testing presentation attack detection (liveness).

Know Your Customer — identity checks regulated businesses must perform during onboarding.

Key facial points — eye corners, nose tip, mouth corners — used to align a face.

A liveness check that works from a single image or short clip without asking the user to act.

Showing the camera something other than a live face: a print, a screen replay or a mask.

GDPR term for sensitive data, including biometric data used to uniquely identify someone.

The similarity score above which two faces are treated as a match; it trades FMR against FNMR.

Create a free account, grab your token and make your first recognition call in minutes.

Start free Talk to our team