Tensor facial recognition partners Herta outshine competitors at US Homeland Security event

Tensor's facial recognition partners, Herta, have managed to outshine competitors at the 2018 Biometric Technology Rally organized by the US Department of Homeland Security, in collaboration with NIST, the company has revealed.

The 2018 Biometric Technology Rally took place in March 2018, at the Maryland Test Facility (MdTF), a DHS affiliated laboratory. Herta came out on the top 3 facial recognition vendors, among the highest scores for efficiency, customer satisfaction and performance.

The rally participants were provided with a gallery of 3,000 images of around 300 people, and each contestant was required to match people's faces one at a time as they were going through a facial recognition checkpoint. The test measured the average transaction time, failure to process an image, percentage of satisfied customers and the effectiveness of providing a positive identification within a 5 and 20 seconds interval.

Herta designed a specific solution for this one to many (1:n) test despite its speciality is many to many (n:m) as in some of its main projects to improve safety in safe cities or sports stadiums.

The scenario simulated an airport control point where the solution compares traveler's passport pictures from the images captured through the CCTV cameras. Herta used its real-time face identification technology to meet the Rally requirements, and it exceeded the average for most of the metrics as well as the 95% successful face identification rate in less than 5 seconds, compared to an average of 65%.

Alex Collado-Castells, CTO at Herta and responsible for the rally deployment, explained that the test was very successful and it helped to prove how the company's technology stands out among other competitors, and to see what governments are looking for when it comes to a checkpoint solution.

"It was an inspiring experience, where we had to work under pressure in order to design a specific solution for a test which reflects a real world use case." Collado-Castells said. "The test generated very useful metrics which are aligned to the current government requirements and needs. Herta performed excellent at efficiency, customer satisfaction and performance, the three main goals of the Rally."

Tensor — experts in deploying high-end facial recognition systems

Tensor's facial recognition solution is perfect for video surveillance and tracking, access control, and commercial marketing management.

The system can dynamically compare images of individuals from incoming video streams, of varying quality, against specific databases and send immediate alerts when positive matches occur. It also quickly manages real-time multi-recognition of faces in highly crowded environments.

The software has an amazingly high level of accuracy and is even able to analyse the 8K resolutions generated by the most high-end CCTV surveillance cameras. Its smart technology has in-built data analytics; detecting and estimating gender, age ranges and ethnicity — there will be more to follow later. The application is also able to conduct 35 million comparisons per second, making it the fastest system in the world.