Skip to content

Face Liveness Detection API

New

Check whether a submitted selfie or capture came from a live person, not a photo, screen or recording. Use the result to decide whether the session continues.

Overview

A face liveness check assesses whether the face in a submitted capture belongs to a person who was actually present. The alternative is a printed photograph, a face on a screen, a replayed video or a mask held up to the camera. The question is narrow on purpose. Liveness does not say who the person is, which is what a face match does. Remote onboarding flows generally use both to establish that a real person is present and that the person matches the identity document on file. Through an API the check runs inside your own capture flow and returns an assessment your workflow can act on. Capture conditions affect what can be assessed, and the detail returned may vary by capture, data source and record. Build a retry path and a review queue into the design.

Category
KYC Services
Coverage
Selfie and short captures collected in your own app or web flow
Input
A selfie image or short capture collected by your own application
Output
A liveness assessment your workflow can act on
Typical users
Lending, insurance, fintech, gig platforms, telecom
Coverage note
Supported capture modes and formats depend on the selected service contract

The overview and the sections after the reference are the FinSour API catalog’s own record for this service, rendered unaltered.

Reference in progress

The request and response reference for this endpoint is not published yet.

Everything below this block is the live catalog record for this service, and it is accurate. The platform’s published API contract does not include this service yet, so there is no request or response detail to document. Rather than fill these sections with a plausible example, the page lists what is missing:

Endpoint URL and method
The host, path and HTTP verb for this service.
Authentication
How a request identifies itself, and where credentials come from.
Request parameters
Field names, types, which are required, and validation rules.
Response schema
The success body, field by field, with a real example.
Error codes
What each failure means and which are worth retrying.
Rate limits
Request ceilings, and what happens at the ceiling.

No sample payload, URL, response body, error code, rate limit or latency figure appears anywhere on this site unless it has been supplied and attributed. Each item in this list is removed only when its own detail arrives; none of them is held back to wait for the rest, and none is ever placed beside a guess in the meantime.

Use cases

  • Lending and NBFC - confirm a live applicant before a remote onboarding step
  • Insurance - live capture on a remote proposal or claim intimation
  • Fintech and PSP - step-up check before a high-risk account action
  • Staffing and gig platforms - shift-start check against shared accounts
  • Logistics and fleet - driver identity check at vehicle handover
  • Telecom - live capture in a remote self-service activation journey

What this service does

  • Photos, screens and masks are what it looks for

    The check assesses whether the capture came from a person actually present. A printed photograph, a face on a screen, a replayed video and a mask held to the camera are the cases it catches. It is the assumption every other remote identity check depends on.

  • How it differs from a face match

    Face match compares two faces and liveness assesses one capture. A borrowed photograph can pass a match while failing liveness. Remote journeys generally run both: one establishes that a real person is present, the other which person.

  • Passive or prompted capture

    A passive check assesses the capture without asking the customer to do anything. A prompted check asks the customer to perform a simple action during capture. Which modes are available depends on the selected service contract.

  • Designed for real capture conditions

    Lighting, compression, an obscured face, an unstable connection and low-end cameras all affect what can be assessed. The detail returned may vary by capture, data source and record. Send a capture that could not be assessed to a retry.

How a check runs

  1. Agree capture modes and volumes

    Start with Get API and describe the journey you are protecting. Available capture modes, supported formats and volumes depend on the selected service contract.

  2. Add capture to your app or web flow

    Collect the selfie or short capture on the customer's device at the point in the journey where the risk sits. Show a clear notice about what you capture and why.

  3. Submit the capture from the live session

    Your application sends the capture in a format and mode the service contract supports, from the same session the customer is in.

  4. Read the liveness assessment

    The service returns an assessment your workflow can act on, along with the detail available for that capture.

  5. Continue, retry or send to review

    Map each outcome to a defined next step. Continue the journey, offer a guided retry where the capture could not be assessed, or route the session to human review.

Questions

What do I send to run a face liveness check?
A capture your own application collects from the customer's device, either a selfie image or a short capture. Use the formats and capture modes the selected service contract supports. The liveness check itself needs no identity document.
What does a face liveness check return?
An assessment of whether the capture came from a live subject, in a form your workflow can act on. The detail returned alongside it may vary by capture, data source and record. Build the flow around the outcome.
How is face liveness different from face match?
Face match compares two faces and reports similarity. Liveness looks at a single capture and assesses whether a live person produced it. A borrowed photograph can pass a match while failing liveness, which is why remote journeys generally run both.
Do I need consent to capture a customer's face?
Yes, a face capture is sensitive personal data and belongs in a consent-led flow. Tell the customer what you capture, what you use it for and how long you keep it. Collect it only for that purpose, and offer an alternative path for anyone who cannot complete a camera step.
Why does one capture assess differently from another?
Lighting, camera quality, compression, connection stability and how much of the face is visible all affect what can be assessed. A capture that could not be assessed is not a failed check. Offer a guided retry before you turn anyone away.
Can face liveness stop identity fraud on its own?
No single check can. Liveness is a signal that raises or lowers confidence in a session. It works best alongside your other controls, including device and application integrity, the channel the capture arrived on, and the human review path you already run.

Catalog entry last updated .