Innovatrics is boasting about its top-three showing in the Department of Homeland Security’s (DHS) latest Biometric Technology Rally. The 2020 Rally was announced last March, and was carried out at the Maryland Test Facility by the DHS Science & Technology Directorate (S&T).
During the Rally, the S&T evaluated the performance of a number of different facial recognition solutions, looking at six systems for acquiring faces and another 13 systems for matching them. Innovatrics did well in both categories, though its results in acquisitions were particularly strong. The company’s solution was the most accurate system for acquiring faces, to the point that it was able to pick out faces even when the subject did not stop to look at the camera. The results indicate that the solution is well-suited to venues with a high volume of foot traffic, or any location that needs to scan a large number of faces in a short amount of time.
The acquisition system was able to detect both masked and unmasked faces, which means that it can be deployed in places that have mask requirements in place to slow the spread of COVID-19. In that regard, Innovatrics’ algorithm was able to identify faces once it had acquired them, standing as one of the three most accurate matching systems in the Rally. That was true in both masked and unmasked situations, and when making matches with low-quality photographs.
“We aim to provide the highest level of accuracy, security and convenience to our users,” said Innovatrics SmartFace CTO Michal Vilagi. “We believe our technology is instrumental in building solutions that elevate both public safety and the user experience.”
Innovatrics updated its SmartFace platform in September to add mask detection capabilities. The company has also performed well in recent rounds of NIST testing.
The S&T conducted the Rally with the help of 582 volunteers from 60 countries all over the world. Corsight delivered the most accurate solution in the unmasked category, and placed second when masks were being worn.
March 3, 2021 – by Eric Weiss