Verify identities
in seconds.

Electronic KYC in one flow: face matching, liveness check and deepfake detection, built on VisionLabs technology.

Mitreka products: workflow, documents, geospatial, IoT, AI, village and vending
OVERVIEW

One flow for
trusted onboarding.

Combine Face Recognition, Liveness Detection and Deepfake Detection in a single verification request, so banks and public services can onboard people remotely with confidence.

99%Face recognition accuracy*
100%Attacks detected in iBeta tests
1 framePassive liveness check
0.1 sSearch across 3M faces

*Based on NIST test results published by VisionLabs. Accuracy depends on image quality and system configuration.

ABOUT THE PARTNER

VisionLabs, world-class biometric technology

VisionLabs

VisionLabs develops face, body and vehicle recognition technology used by government, finance, telecom and retail clients in many countries. Its software runs on more than 1.7 million cameras and processes over 1.5 billion events every month.

14 yearsComputer vision expertise
300+Clients across sectors
1.7MCameras worldwide
9Countries with offices
Mitreka's role as product implementor
Solution designDefine the verification flow around your process.
Installation & setupOn-premise deployment in your environment.
System integrationConnected to core banking or government systems.
Local supportTechnical support and maintenance by the Mitreka team.
THREE MODULES

Everything eKYC needs, in one flow

Choose a module to jump to its details.

Face Recognition
MODULE 01

Face Recognition

Match the selfie with the ID photo or a reference database.

Read more
Liveness Detection
MODULE 02

Liveness Detection

Confirm a real person is present and block spoofing attacks.

Read more
Deepfake Detection
MODULE 03

Deepfake Detection

Detect face swaps, synthetic faces and reenactment.

Read more
Jump to module
MODULE 01 · FACE RECOGNITION

Know the face. Trust the identity.

Biometric face matching for eKYC onboarding, secure login and identity verification, built on VisionLabs LUNA PLATFORM 5.

99%Recognition accuracy*
120MFaces per database
10,000Requests per second
0.1 sSearch across 3M faces

*Based on NIST test results published by VisionLabs. Accuracy depends on image quality and system configuration.

KEY FEATURES

1:1, 1:N and N:N matching

Verify a selfie against an ID photo, or identify a person across millions of records.

Robust to appearance changes

Stays accurate when users wear masks or glasses, or have a beard.

Image quality & ICAO checks

Checks photos against ISO/IEC 19794-5 and ICAO recommendations for identity documents.

Facial attributes

Estimates age, gender, emotions and accessories such as glasses or headwear.

Lists, events & notifications

Manage watch lists in the LUNA CLEMENTINE web interface with role-based access.

Anti-spoofing modules

Add Liveness and Deepfake Detection to the same verification request.

INDEPENDENTLY TESTED
NIST FRTE#1 in 4 of 6 verification and 6 of 8 identification categories.
ISO 9001:2015Certified VisionLabs quality management system.
ISO/IEC 19794-5 · ICAOPhoto compliance checks for identity documents.
MODULE 02 · LIVENESS DETECTION

Make sure it is a real person.

Detects printed photos, screen replays and 3D masks from a single selfie frame, with no extra gestures needed.

iBETA · ISO/IEC 30107-3Level 3Also passed Level 1 & 2
100%Attacks detected in iBeta tests
3,300Attack attempts (L1 + L2)
1 framePassive check
ATTACKS PREVENTED
Printed photo illustration
Printed photoA photo of another person printed on paper.
Paper mask illustration
Paper maskA face cut out from a printed photo.
Screen replay illustration
Screen replayA photo or video shown on a screen.
3D mask illustration
3D maskA volumetric face or head mask.
KEY FEATURES

Passive one-shot liveness

One selfie is enough. No blinking, turning or smiling.

Active liveness option

Add guided camera interaction when your risk policy requires it.

Same image as verification

Runs on the frame used for face matching, with no extra step.

Depth-channel support

Uses depth data from compatible cameras as an additional check.

Pairs with Deepfake Detection

Protects against physical spoofs and AI-generated faces.

Server version

Backend version available; test data for other versions on request.

INDEPENDENTLY TESTED BY IBETA
iBeta PAD Level 21,500 attack attempts, 100% detected.
iBeta PAD Level 11,800 attack attempts, 100% detected.
MODULE 03 · DEEPFAKE DETECTION

Stop fake faces made by AI.

Identifies AI-manipulated faces, from face swaps and synthetic faces to reenactment, before they enter your eKYC flow.

99%Overall detection accuracy*
97.5%Real people correctly accepted
97.3%Fake faces identified
3 typesDeepfake attacks recognised

*Based on VisionLabs algorithm testing. Accuracy depends on image quality and system configuration.

THREATS RECOGNISED
Face swap illustration
Face swapA face from one image is transferred onto another person's photo or video.
Face synthesis illustration
Face synthesisA realistic face of someone who never existed, fully generated by AI.
Face reenactment illustration
Face reenactmentAnother person's expressions and movements drive a real face.
KEY FEATURES

Real-or-fake score

Shows whether an image is likely genuine, with a confidence level.

Virtual camera attacks

Helps detect deepfakes injected into the video stream.

Same image as verification

Runs on the frame already used for face matching.

Pairs with Liveness

Full coverage of physical and digital attacks.

Part of LUNA PLATFORM 5

Licensed as a module and called through the same API.

Audit-ready results

Every result is logged as an event for fraud or compliance review.

KEY FEATURES

Secure onboarding, without the friction

ID photo matching

Compare a selfie with the photo on an e-KTP or a reference record.

Passive liveness

One selfie is enough, with no blinking or head turns.

Deepfake protection

Stops AI-generated and manipulated faces.

Image quality checks

Checks photos against ISO/IEC 19794-5 and ICAO recommendations.

One request, one decision

Run all three checks together through a single API.

On-premise deployment

Runs in your own environment, so biometric data stays with you.

HOW IT WORKS

From selfie to verified identity

All checks run automatically in seconds and return one result to your system.

01CaptureThe user takes a selfie and a photo of their ID.
02Check livenessConfirm a real, present person.
03Detect deepfakesRule out swapped or synthetic faces.
04Match & decideCompare with the reference and return a decision.
CONTACT US

Bring trusted eKYC to your onboarding

Talk to the Mitreka team about your requirements and integration.

Contact Us