Unissey aces iBeta Level 2 biometric PAD test on both mobile and laptop
French facial biometrics firm Unissey claims a world first after its passive liveness detection algorithms pass the iBeta Presentation Attack Detection Evaluation Level 2 on both mobile and laptop formats. The results from testing in December and January show that all attacks were blocked.
U.S.-based iBeta provides software testing including those compliant with ISO 30107-3 Biometric Presentation Attack Detection (PAD) Standard Level 1&2. Level 2 is significantly more complex. Tests involve attacks such as silicone and latex masks, video manipulation software and various printed photo types to try and fool the algorithms for human liveness detection in identity authentication.
Unissey chose a Samsung Galaxy S20 running Android 10 for its smartphone option and an HP laptop running Windows 10 Home, considering them to be universal. No attacks were successful on either device, an APCER (Attack Presentation Classification Error Rate) of zero percent. Each device was tested 750 times.
The BPCER (Bona Fide Presentation Classification Error Rate), the false reject rate on a first attempt at attack, had a result of 0 percent on the mobile and 2 percent on the laptop.
The news comes after Unissey recently notching up an 18 percent improvement in its NIST Face Recognition Vendors Test (FRVT) scores for its face biometrics algorithms.