Skip to main content

Face Recognition & Liveness Verification

This document explains how Face Recognition and Liveness Verification work from a user perspective, what is required during the process, and how captured data is used.

1. What is Face Recognition?​

Face Recognition is a security process used to confirm that:

  1. You are a real, present human being.
  2. Your face matches the reference image you previously provided (for example, a profile photo).

This helps prevent unauthorized access using photos, videos, or impersonation attempts.

2. Why is Liveness Verification required?​

A face image alone is not sufficient to guarantee security.

Liveness Verification (also called Proof of Life) ensures that:

  • The interaction is happening in real time
  • A real person is physically present in front of the camera
  • The system is not being tricked by:
    • Printed photos
    • Screens or replays
    • Pre-recorded videos
    • Masks or deepfake attempts

3. How the verification process works​

Step 1: Reference image validation​

The system checks whether the reference image contains at least one human face.

If no face is detected, the process cannot continue.

Step 2: Live camera verification​

You will be asked to enable your camera and follow on-screen instructions.

These instructions may include:

  • Looking left or right
  • Looking up or down
  • Closing your eyes
  • Opening your mouth

The sequence is randomized to prevent spoofing.

Step 3: Proof of Life confirmation​

While you perform the requested actions, the system verifies that:

  • Your face is real and moving naturally
  • Only one face is visible
  • Your face remains within the guided area on the screen
  • Gestures are performed correctly and within the allowed time

If the liveness challenge is not completed successfully, the verification will fail.

Step 4: Face comparison​

Once liveness is confirmed, the system:

  • Captures high-quality frames from the live session
  • Extracts facial features
  • Compares them with the reference image

If a sufficient match is found, the verification succeeds.

4. Enabling Face Recognition and Liveness Detection​

Face Recognition and Liveness Detection can be enabled from the Tenant Configuration panel under Security & Access.

These settings allow you to control how user identity is verified during authentication.

Step 1: Enable Face Recognition Authentication​

To require face recognition for user identity validation:

  1. Go to Backoffice → Tenant Configuration
  2. Open Security & Access
  3. Enable Face Recognition Authentication

Once enabled, users will be required to verify their identity using face recognition when validating a Keepic.

After Face Recognition is enabled, you can activate Liveness Detection for additional security.

  1. Make sure Face Recognition Authentication is already enabled
  2. Toggle Use Liveness Detection
  3. Save the changes

When this option is active, users must complete a live gesture challenge (Proof of Life) to confirm they are physically present and not using a photo or video.

For maximum security, we strongly recommend enabling both:

  • Face Recognition Authentication
  • Liveness Detection

This combination provides protection against impersonation, replay attacks, and spoofing attempts while maintaining a smooth user experience.

5. Possible results​

After the process, one of the following outcomes is returned:

  • Success
    Your identity was verified successfully.

  • Warning
    The match is acceptable but has lower confidence. Additional backend validation may occur.

  • Fail
    The system could not confirm your identity or liveness.

6. Common reasons for failure​

Verification may fail if:

  • The camera is blocked or blurry
  • Multiple faces appear on the screen
  • The face leaves the guided area
  • Required gestures are not completed within the allowed time
  • A static image or replay attempt is detected

7. Data Privacy​

Does this process store any biometrics?​

No. This process does not generate or store any biometric data (such as biometric templates or patterns).

It only validates the presence of a real person in front of the camera and compares the live session against the user-provided reference image (Keepic).
All processing happens on the fly, entirely in the memory of the user’s device, without generating or storing any biometric data locally or remotely.

Does this process send any personal data to servers?​

No. Video frames and images are processed exclusively in memory on the user’s device.

Nothing is:

  • Sent to external servers
  • Stored locally
  • Persisted after the verification process ends

No personal or biometric data is exposed outside the user’s device during face recognition or liveness verification.

The only information sent to the server is the final outcome of the verification process.

Which data is generated?​

The face recognition process only generates and shares generic, non-biometric validation data, such as:

  • The liveness challenge sequence executed (face movements)
  • Execution and duration times
  • The verification result (pass or fail)
  • Face similarity score

This data does not allow biometric reconstruction and is used strictly for validation and audit purposes.

8. Security and anti-spoofing measures​

To protect users, the system includes safeguards such as:

  • Detection of unnatural lack of movement
  • Blink monitoring
  • Frame repetition detection
  • Head pose variability checks
  • Multi-face prevention

These measures help ensure that only real users can pass verification.

9. What users need to do​

For best results:

  • Use a well-lit environment
  • Avoid backlighting
  • Remove hats, masks, or sunglasses
  • Hold the device steady
  • Follow on-screen instructions calmly

10. Summary​

Face Recognition and Liveness Verification work together to provide a secure, privacy-respecting way to confirm identity.

The process is fast, non-invasive, and designed to protect both users and the platform from fraud or impersonation.

If verification fails, simply try again by following the guidance provided on the screen.