DHS Publishes RIVR Selfie-to-ID Verification Results; Paravision Cites Top Performance

The Department of Homeland Security has published results for the Remote Identity Validation Rally Selfie Match to Document track, a government-run benchmark evaluating how well facial recognition systems match a live selfie to a photo on a U.S. state-issued ID.

The testing, which follows DHS’s launch of the RIVR program to benchmark remote identity verification systems, evaluated 16 vendor systems using data collected from 1,632 volunteers across multiple state IDs and three smartphone models. Results are published by the Maryland Test Facility with vendor systems anonymized as MTDS 1 through MTDS 16.

The SMTD track measures selfie and document failure-to-extract rates, false non-match rate, and false match rate. DHS set a verification threshold intended to achieve a false match rate below one in 10,000.

Paravision announced that 31 percent of tested systems met RIVR goals across all reported metrics, while 63 percent met minimum thresholds and 37 percent failed to meet at least one goal or threshold. The company identified its system as MTDS 12 and said it was the only system to meet all goals and the look-alike imposter false match rate goal.

Paravision has previously highlighted results such as its DHS face liveness test performance in RIVTD Track 3. Other DHS benchmarking outcomes include IDEMIA’s remote identity validation results from RIVTD testing.

RIVR SMTD results provide an apples-to-apples comparison of selfie-to-ID verification performance and can influence vendor selection, procurement decisions, and technical requirements for remote identity verification programs across government and private-sector deployments.

The Maryland Test Facility notes that companies can be contacted through DHS to map anonymized system aliases to specific vendors. Results are reported as worst-case performance by device and document state.

Sources: Maryland Test Facility, RIVR SMTD Results PDF, Paravision

–

By the ID Tech Editorial Team