Selfie liveness detection
How it works
Persona collects active and passive signals from the user, their selfie, and the device they’re using.
Key features
Deploy your multi-layered approach with proprietary signals, the latest models, and dynamic friction
Alignment with EU standards for injection attack detection Toggle description visibility
Persona’s multi-layered solution aligns with CEN/TS 18099:2024 standards.
Industry-leading library of proprietary signals Toggle description visibility
Collect the widest array of signals to fully understand the context surrounding selfie submissions.
Frequently updated analysis models Toggle description visibility
Automatically use the latest proprietary models, purpose-built by Persona’s specialized research and fraud teams.
Multi-frame analysis Toggle description visibility
Capture and analyze multiple selfie frames to enhance accuracy and reduce fraud.
Compromised hardware detection Toggle description visibility
Flag emulators, rooted devices, and other types of compromised hardware fraudsters use to perform injection attacks.
Scaled attack protection Toggle description visibility
Link selfies with similar characteristics back to the same threat actors or fraud rings so you can block scaled attacks.
Configurable settings and checks Toggle description visibility
Adjust image quality requirements, face obstruction detection, retry attempts, and more to meet nuanced compliance, fraud, and business requirements.
Equitable across demographics and devices Toggle description visibility
No instances of material bias across sex, skin tone, and age with streamlined UX on all modern operating systems and devices.
Auto-capture Toggle description visibility
Automatically capture photos at the ideal moment to decrease user error.
User guidance Toggle description visibility
Users receive gesture guidance to decrease user error and provide valuable fraud signals.
Automatically catch visual inconsistencies indicative of deepfakes and other face spoofs — even if they’re invisible to the human eye.
Automatically detect presentation attacks in compliance with ISO/IEC 30107-3 Presentation Attack Detection standards.
Automatically detect sophisticated injection attacks, even if fraudsters use stolen selfies of real people.
Quickly recognize fraud patterns and block fraud rings before they can scale.
Arm manual reviewers with liveness detection signals and analysis results so they can make more accurate decisions with more confidence.
Add friction to deter fraudsters while ensuring good users convert seamlessly based on real time liveness detection results.
Extensively tested and certified by third parties
Persona selfie recognition has achieved superior results for face matching, liveness, and age assurance.
Our models achieved a false non-match rate (FNMR) of 0.0018 at a false match rate (FMR) threshold of 0.000001 as reported by NIST.
Based on separate evaluations by NIST and the Age Check Certification Scheme (ACCS).
iBeta found Persona to be compliant with both Level 1 and Level 2 of the ISO/IEC 30107-3 Presentation Attack Detection Standard with a measured Attack Presentation Classification Error Rate (APCER) of 0%.
Persona’s age estimation models achieved an MAE of 1.4 years for minors under 18 based on testing by the Age Check Certification Scheme (ACCS).
Persona’s age estimation was included in the overall concept positively evaluated by the Kommission für Jugendmedienschutz (KJM).
