The U.S. Department of Homeland Security’s Remote Identity Validation Rally (RIVR) has completed evaluations across selfie-to-document matching and liveness detection, with all testing conducted on consumer smartphones, establishing performance benchmarks directly relevant to mobile identity verification deployments.
In Track 1, which tested systems’ ability to match a live selfie captured on a smartphone to a government-issued ID document, 16 vendors were evaluated using 1,632 volunteers at the Maryland Test Facility. Participants used three smartphone models: iPhone 14, Google Pixel 7, and Samsung Galaxy S22. Five systems met all DHS high-performance goals for failure-to-extract rate, false match rate, and false non-match rate.
Four of the five qualifying vendors have publicly identified themselves: Paravision, which reported the lowest false match rate with 26 false matches across nearly 8 million impostor comparisons; Aware; Incode; and Shufti. A fifth vendor met the goals but has not disclosed its identity. DHS anonymizes all vendors in published results.
Paravision was the only system to also meet the rally’s most demanding benchmark: maintaining a false match rate below 0.01 percent when tested against demographically similar impostors, a scenario designed to simulate look-alike fraud. For the median system, false match rates were approximately 11 times higher for demographically similar pairs than for randomly selected impostors.
Today, Aware announced results from Track 3, which evaluates presentation attack detection on mobile devices. The company said its Intelligent Liveness system achieved zero successful attacks across attack classes A and B in the active liveness evaluation, which uses live human participants rather than pre-collected images. Aware also reported top-three rankings in user satisfaction on both iOS and Android and top-two rankings in processing speed, metrics that speak directly to mobile deployment viability.
The smartphone-native testing methodology distinguishes RIVR from laboratory evaluations that use controlled imaging equipment. By requiring vendors to perform on the same consumer devices their systems would encounter in production, the results offer a more realistic assessment of how mobile identity verification performs under real-world conditions, including variable lighting, camera quality differences across manufacturers, and device-specific processing constraints.
Sources: Aware, Maryland Test Facility
–
By the Mobile ID World Editorial Team
