Innovatrics Halves Error Rate in Latest FRTE Performance

Innovatrics has made significant strides with its latest facial recognition algorithm, achieving a top-tier ranking in the National Institute of Standards and Technology (NIST) Face Recognition Technology Evaluation (FRTE) 1:N benchmark.

The new algorithm halved its error rate, giving Innovatrics a top-13 placement out of 152 global vendors. Innovatrics is also among the few vendors with top-ranking algorithms for iris, fingerprints, and facial recognition.

The advancements in the Innovatrics facial recognition algorithm include an improved detection model, a revised approach to final score computation, and an enhanced re-ranking algorithm. These enhancements have collectively propelled the algorithm to the top 10 percent of the global leaderboards, showcasing substantial progress in reducing identification error rates.

Jakub Sochor, head of Innovatrics R&D team, highlighted the focus on two critical benchmarks: VISA-BORDER and MUGSHOT-MUGSHOT. In the VISA-BORDER category, the algorithm compares faces captured during border crossings with visa application photos.

The MUGSHOT-MUGSHOT category involves matching faces within galleries of 1.6 million and 12 million photos, where Innovatrics’ algorithm achieved notable gains in both scenarios.

The NIST FRTE 1:N benchmark is an extension of the previous FRVT tests and evaluates the accuracy of face recognition technology. Innovatrics’ latest results in this benchmark show improvements, with error rates much lower than in earlier versions. These improvements will be incorporated into production in the coming months.

–

July 12, 2024 – by Ali Nassar-Smith