Compliance

At Mobai, we know that quality and standards compliance are essential when choosing a biometric vendor. That’s why we are committed to transparent testing, continuous improvement, and achieving relevant certifications — building reliable, proven biometric solutions our partners can trust.

Mobai Liveness and Presentation Attack Detection (PAD) Achieves Perfect Results in Independent Testing
Mobai’s face Liveness and Presentation Attack Detection (PAD) system has been independently evaluated by the Swiss Center for Biometrics Research and Testing (SCBRT), a FIDO Alliance Accredited Biometrics Laboratory (ABL) and part of the Biometrics Security and Privacy group at the Idiap Research Institute.

The evaluation was conducted in accordance with ISO/IEC 30107-3, the internationally recognized standard for biometric presentation attack detection protocols. Mobai’s solution achieved a perfect score, with:
- Bona Fide Presentation Classification Error Rate (BPCER): 0%
- Attack Presentation Classification Error Rate (APCER): 0%

Testing covered both Level A and Level B attack instruments, representing a range of presentation attack sophistication:
- Level A: Lower-quality printed images or digital displays
- Level B: More advanced attacks such as videos and 3D masks

Mobai’s system demonstrated robust resilience across these attack types. The evaluation used a 3-frame fusion strategy within 1 second, reflecting realistic usage scenarios and minimal latency.
‍
In addition to ISO/IEC 30107-3 compliance, Mobai’s approach aligns with the broader requirements and principles of ISO/IEC 19989, which governs the security evaluation of biometric systems, ensuring privacy, integrity, and trustworthiness.
‍
These results confirm Mobai’s solution as a highly secure and reliable biometric PAD system, suitable for deployment in demanding environments where trust and compliance are critical.

Mobai Face Verification Technology Validated by NIST
Mobai’s proprietary face verification algorithm, developed in-house using industry-standard deep learning frameworks, has been independently evaluated by the National Institute of Standards and Technology (NIST) as part of its Face Recognition Vendor Test (FRVT). The results demonstrate strong performance across key verification scenarios and confirm alignment with BITS security requirements.

NIST conducts rigorous evaluations on undisclosed datasets comprising millions of facial images collected under diverse real-world conditions, including visa applications, mugshots, border crossings, and kiosk environments. These images span over 100 nationalities and age groups, making NIST’s FRVT one of the most authoritative benchmarks for biometric performance worldwide.

A particularly relevant evaluation for identity verification use cases—such as remote onboarding—is the Visa vs. Border photograph comparison. This test closely mirrors the real-world challenge of comparing a passport-style photo to a live selfie. In this scenario, Mobai’s algorithm demonstrated a False Non-Match Rate (FNMR) of 0.0073 at a False Match Rate (FMR) of 1e-5, and 0.0093 FNMR at FMR = 1e-6, offering both high security and user experience.

For Mugshot images:
- FNMR = 0.0047 at FMR = 1e-5
- FNMR = 0.0075 at FMR = 1e-6

For Visa images:
- FNMR = 0.019 at FMR = 1e-5
- FNMR = 0.0219 at FMR = 1e-6

These results meet or exceed BITS' requirement of FNMR < 0.02 at FMR = 1e-5 for all major test cases, with most tests also compliant at the more stringent FMR = 1e-6.Based on NIST testing assumptions and conversion models, Mobai’s visa–border matching scenario yields an accuracy of 99.9999%, significantly exceeding the 99% success rate required by BITS for confidence-level algorithms.