Results from the third phase of the Department of Homeland Security’s Remote Identity Validation Rally show significant variability in how well biometric liveness detection systems on the market can resist spoofing attacks, with only a handful of entries meeting the evaluation’s performance goals across its active and passive presentation-attack-detection tracks.
The evaluation, conducted by the DHS Science and Technology Directorate at the Maryland Test Facility, assessed six active and twelve passive PAD systems. In the passive track, Paravision’s entry was the only submission to meet the RIVR goal in one category and the threshold in all others, posting a bona fide presentation classification error rate below 0.5 percent and an attack presentation classification error rate of 1.7 percent. IDEMIA Public Security announced that its active-track entry met all RIVR performance thresholds, including satisfying requirements for security, user experience, and transaction time.
Identy.io said its PAD-A2 system was the only active-track entry to record zero attack acceptance across three attack classes tested under independent scoring conditions. The company also said it posted the fastest transaction times and the highest user satisfaction scores among the active-track field. Identy.io framed its performance around a fully passive approach that does not require users to turn their heads, follow a moving prompt or speak into the camera.
The broader RIVR Phase 3 picture is mixed. Among the twelve passive systems, only three met the bona fide classification error rate goal and five more met the threshold, while just one met the attack acceptance threshold. Most active systems also fell short on at least one metric. The results follow Phase 2, which found document validation performance across the field to be poor, and Phase 1, which showed selfie-to-document matching to be more reliable overall.
Identy.io’s announcement extends a run of benchmark-focused claims from the company. In February it claimed a top-three result in the first NIST fingerprint identification benchmark, and earlier the same month it announced expansion into Kenya and Nigeria. Independent evaluations have become a standard competitive tool across the sector, with vendors including ROC citing top rankings in NIST latent fingerprint benchmarks to differentiate their offerings.
Sources: FinancialContent, Maryland Test Facility, IDEMIA
–
By the ID Tech Editorial Team
