Oz Liveness

Main article: Biometric identification technologies

2021: Compliance with ISO 30107-3

The Oz Liveness Oz Forensics passive detection solution biometric attacks passed the ISO 30107-3 standard test. This became known on April 12, 2021. The check took place in the laboratory of iBeta accredited to. National Institute of Standards and Technology (NIST)

With the transition to remote user identification around the world, fraudsters specializing in the use of biometric and deapfake attacks have intensified. Criminals, as a rule, resort to spoofing - this is a demonstration of photos or videos of another person from a laptop or tablet for remote identification. They can also use a mask.

The Oz Liveness solution was developed to determine whether a living person is in the frame. Testing for iOS and Android platforms in the laboratory iBeta was carried out using the following artifacts: 2D photo on matte paper with cutouts, curved on the face, mask with an eye cut, photos and videos from a laptop or iPad, 3D paper mask made manually. In some attacks, hats and glasses were added. Of the 300 original Liveness tests that were performed with a real face, 299 completed successfully. False Oz Liveness error was less than 1%. At the same time, none of the 1000 attacks, that is, were missed. the accuracy of attack detection was 100%.

Our deep machine learning based algorithms check video frames and track dozens of parameters (glare and reflections, micro movements, pulse, etc.) We've trained the system on tens of thousands of attacks. We believe it is important to work with 3D mask manufacturers and are constantly in search of new samples. It was this approach that allowed our Oz Liveness decision to pass ISO 30107 certification in a NIST-accredited laboratory iBeta with a result of 100%, "said Artem Gerasimov, CEO of Oz Forensics.

The Oz Liveness algorithm is passive, that is, it does not imply any active action, the user only needs to look into the camera. One frame is enough, thereby simplifying the transmission and processing speed of 1 process per second. Conversion is not reduced because no additional action is needed.

In addition, Oz Liveness reduces the risks of remote user identification and transaction confirmation, reduces the cost biometric authentications of payment by the person ACS and in (access control system), as the need for infrared cameras does not fall. Oz Liveness ensures client security personal data during exchange and. storage of information

The mobile SDK and Oz Liveness API are used by dozens of customers around the world in Africa, Asia, Europe, the CIS - these are banks, microcredit organizations, civil aviation companies, telecom operators.

2020: Announcement of technology for customer identification and verification

Oz Forensics Oz Forensics, a resident of Skolkovo, announced on June 19, 2020 the creation of a technology for identifying and verifying customers, which is used to open bank accounts remotely.

that Coronavirus has struck the whole world and plunged economies many countries into the deepest crisis, oddly enough, has caused some positive changes. The epidemic accelerated the transition to a remote format of industries that were already striving for it. So, in the conditions of quarantine, remote communication of banks communication with customers became the only possible. And from the end of April 2020 CENTRAL BANK OF THE RUSSIAN FEDERATION , he allowed to open socially significant accounts without a personal presence in the office. This immediately caused a flurry of biometric, as attacks a result of which the demand for technologies for identification and verification of the client/user remotely using biometric on the recognitions face and the definition of "living/non-living" increased. noted Artem Gerasimov, founder and CEO of Oz Forenzics

Oz Liveness 2020

Oz Forensics software helps to identify and verify a customer through the bank's mobile application: compare a person's face with a person on a passport, validate a passport and determine a living person in a frame. It is the definition of Liveness that allows you to resist biometric attacks based on artificial intelligence, for example, deepfake (video synthesis based on photos). An attacker can frame another person's video or photo, use a mask and get a loan, open a fake account and account, or receive another person's allowance when everyone is sitting home due to quarantine. The Oz Forensiicsc technology with an accuracy of 99.87% identifies such biometric forgery. The system recognizes gender, age and nationality, selecting more accurate verification parameters for a certain group of people and is resistant to the presence of a beard and other facial changes.

Liveness technology - for 2020 is relatively new for the market, it is no more than 10 years old, while the technology for recognizing facial biometrics and facial comparison has its history since the 1970s. In 1950, Alan Turing proposed the Turing test, which determines human behavior using algorithms, one of its types is the well-known Captcha - a proposal to confirm that you are not a robot on sites.

As of June 2020, one of the most accurate biometrics tests is NIST FRVT, but for certification of Liveness algorithms for compliance with the ISO standard, there is only one commercial laboratory accredited by NIST, which conducts paid tests using its details, and it is located in the United States.

Liveness technology is more difficult to replicate because it requires serious investment and expertise. For example, to train a biometric face comparison system, you can use databases of high-quality photographs that are in the public domain, while to define a "living - not living person," the system needs to be trained on specially created biometric attacks. Such bases are not available, and such attacks must be reproduced independently, which leads to an increase in the cost of technology and the difficulty of its repetition.

Oz Liveness technology using artificial intelligence allows you to recognize and identify a living person in a video frame and cut off biometric forgery. The system analyzes a variety of parameters ranging from pulse to specific light properties reflected from the face. For the client himself, this process is as follows: the system asks the person to perform one of the selected actions, for example: smile, turn his head, bring his face closer or away or otherwise, then she analyzes the video and issues a verdict: Liveness passed or not. It takes no more than 3 seconds.

One of the first cases for remote verification of the user for wallets Russia in was implemented in. In payment service provider QIWI other countries where remote identification was allowed at the legislative level even earlier, technology is in demand among banks MFIs, payment systems, state remote identification providers.

2019: Integration into the mobile application "Altyn Bank"

In 2019, for Altyn Bank in Kazakhstan, a digital bank without offices for customers, introduced Oz Liveness technology into its mobile application. The solution allows you to determine the presence or absence of a living person in the frame (video) by asking him to make a short video. The process became fully automated, and the client verification time was reduced to 3 seconds, while the risks of opening a fake account did not increase.