A Belarusian team has developed a neural-biometrics access management system called SVORT, which is much in demand outside Belarus. The system verifies real users using their faces. The only equipment needed is a webcam.
Here is how it works. A neural network learns to transform a scanned face into static code, which can be used as an authenticator for granting access or as a private key for multi-factor authentication systems. The system also protects users from spoofing attacks employing photos, videos and masks. No special equipment is required. The system can be integrated into any existing online or offline security infrastructure and is compatible with any kind of webcam, turnstile or electronic lock.
SVORT’S CEO Eugene Mintsevich noticed a worrying problem while he was doing business: «The market was being affected by the problem of safe authentication, and businesses were suffering irrecoverable losses due to the existing software solutions being imperfect. Currently, there are two kinds of multi-factor authentication. The first is OTP (authentication through SMS, e-mail, etc.), which is convenient but insecure. The second is U2F, which stands for cryptographic authentication using a separate device. This is secure but inconvenient.»
In December 2018, the idea occurred to create a software authenticator that would be safer than OTP and more convenient than U2F. By March, the MVP had been completed, and the team started to look for investors. In April, they presented a pitch deck to Minsk Business Angels Band, yet the local angels did not consider the product to be marketable and scalable. Nevertheless, in May the team attracted their first $200K in the form of funding from American and Kazakhstani investors.
SVORT’s team continued to work hard on the project over the summer. In the process of their market research, it was decided to improve the authenticator by adding biometrics as an extra security factor.
Already, by then, the embedded biometrics were different from FaceID and similar solutions as they enabled users to use the biometrics on any gadget after registering once on a single device. Notably, neither references nor a photo or any other sensitive data are stored on users’ devices or servers.
Eugene continues the story: «After raising funds, we decided to focus on business development. Our co-founder, the Azerbaijani Geray Yusifov, helped with this. His large network helped us open many business doors. During business trips to the USA, Azerbaijan, and Kazakhstan, we came to understand that local markets needed such technologies. This led to the project’s transformation. From an authenticator, it turned into a neural-biometrics access control system. It turned out that developed technologies are in demand in many areas, yet potential customers need certain functions more than the authenticator itself.»
Now the system comprises 3 main and 2 additional blocks. By combining these blocks, it is possible to perform any case of identification, authentication or verification.
Eugene is happy to report that the system has attracted significant early interest: «Banks are interested in the anonymous biometrics system that I mentioned above. Others are interested in using the authenticator as an access management system. One person is impressed with the system speed and offered to utilize it on turnstiles. We have drawn up agreements on a few pilot projects with three banks from Kazakhstan and Azerbaijan. Employees of the Kazakhstan National Fund are already passing through to the central office using our system rather than cards. In 2020, we plan to install it on all 20 floors of their building.»
He also specified that the main feature of the new system is that there is no need to change the existing security infrastructure. The SVORT neural network learns to recognize the faces of employees while they continue to use the standard access system, gradually replacing traditional keys.
The team is currently actively preparing to attract funding for the planned scaling in the seed-round.