What is face search?

Face search compares a face with a collection of existing portraits to find possible matches.

Also called one-to-many face matching or 1:N matching, it helps recognize returning users and identify possible duplicate accounts, even when document numbers or personal details differ.

Where you would use it

Duplicate-account detection

Find possible links to existing accounts when a person registers with a different credential or identity details.

Returning users

Recognize possible returning users even after a passport renewal or a change from a passport to an ID card.

Repeated registrations

Flag possible repeat enrollments for review when access or benefits are intended to be limited to one account per person.

Verification and identification

One-to-one face verification compares a person with one reference portrait, such as the portrait in the passport they are presenting.

One-to-many matching searches a reference collection, often called a gallery. It asks whether a similar face is already present and which existing records may belong to that person.

This adds a different signal to onboarding: a credential may be genuine and belong to its presenter, while that person already has an account under another credential.

How a face search works

  1. Prepare the reference gallery

    Associate reference portraits with your existing user records. The gallery defines who a search can find.

  2. Compare a new portrait

    Extract a representation of facial features from the search image and compare it with the gallery's representations.

  3. Evaluate possible matches

    Use similarity thresholds to select candidates, then assess them alongside credential details and account history before deciding what to do.

Duplicate accounts across credentials

Matching only names or document numbers can miss a returning person. Someone may renew a passport, use an ID card instead, change their name, or present a digital identity credential.

Face matching provides another way to connect those records. It can support duplicate-account review, recognition of returning users and investigation of repeated attempts to claim account-specific benefits.

A possible match is a reason to examine the records. Multiple credentials or accounts can have legitimate explanations.

Interpreting search results

A search returns ranked candidate records and similarity scores. Your integration applies an acceptance threshold and evaluates the candidates against your account records. A candidate is a possible match, not an identity established independently of the gallery.

Thresholds balance missed matches and false positives. Gallery size and image quality affect search performance, so evaluate the system with data representative of your users and intended collection size.

Review possible duplicates before taking action against an account. A search with no match also does not establish that the person has never used the service; the relevant record may be absent or the comparison may miss it.

NIST research on one-to-many face identification

Trusted portraits and live capture

A portrait from a verified passport or ID chip, or a digital credential containing a signed portrait, provides an issuer-backed reference for the search.

Searching that portrait connects credential records. To connect the current presenter to the credential, combine the search with selfie-to-portrait verification and liveness detection.

Portrait quality and changes in appearance over time matter. A renewed document may provide a more recent image, even when it belongs to someone already in the gallery.

Managing the reference gallery

Face search requires reference data to remain available for future comparisons. Decide which users belong in the gallery, why they are included, who can search it and when their records should be removed.

Gallery retention is a separate requirement from a short-lived verification session. Agree the storage, access and deletion arrangements for your deployment, and explain the purpose of matching to your users.

See our Data practices for standard verification-session handling and contact us to discuss gallery requirements.

Face search with PassportReader

PassportReader offers face search through the Tailored plan. Combine verified credential portraits with your existing user records to help recognize returning people and detect possible duplicates.

Gallery setup, deployment and acceptance policies depend on your use case. Talk to us about your integration or compare the pricing plans.