Skip to main content

Biometric Authentication

Authentication lets an existing Customer prove it is them using their face — or their palm — no document needed. It is a fast 1:1 check against the biometrics already enrolled for that Customer.

When to use it​

Approving a high-value transaction, unlocking a sensitive action, or re-confirming identity at login (step-up authentication). The Company already knows which Customer is being authenticated and passes that Customer to the session.

How it works​

  1. Face capture & liveness — the Customer takes a selfie; liveness confirms a live person.
  2. 1:1 match — the selfie is compared against the face stored for that specific Customer.
  3. Decision — the match score and liveness result combine into an outcome returned to the integrator.

Because the target Customer is known in advance, this is a 1:1 comparison (this face vs that Customer), not a 1:N search across the database.

Outcomes​

OutcomeMeaning
AuthenticatedThe face matched the Customer and liveness passed.
RejectedThe face did not match, or liveness/anti-spoofing failed.

Palm authentication​

Optional feature

Requires the palm verification add-on and a Customer who enrolled both palms during onboarding.

A palm authentication workflow works the same way with a different modality: the Customer shows either palm to the camera, the platform identifies the hand, checks liveness and injection, and compares the palm 1:1 against the stored template of the same hand. The result is match or no_match, judged against the palm matching score Trust Factor.

  • If the Customer has no palms enrolled, the session cannot be started — the API returns an error before any capture, so your app can offer face authentication instead.
  • Face and palm can be combined in one workflow for a two-factor biometric check; the back office then shows both comparisons in the Matching tab.

See Palm Verification for enrolment, Trust Factors and limits.

Notes​

  • No document capture and no OCR — authentication is purely biometric.
  • Anti-spoofing (liveness, deepfake, injection detection) applies here exactly as in onboarding.
  • The accuracy of the 1:1 face comparison technology is independently benchmarked by NIST in the FRTE 1:1 Verification evaluation. The Innovatrics algorithm is listed as innovatrics-017; its row shows the false non-match rates at fixed false match rates across the evaluation datasets.
  • To identify an unknown person by face instead of confirming a known one, use Biometric Identification.

See also​