VisionLabs VisnLabs

The company "VisnLabs" LLC was founded in 2012. It has the official status of a resident of the innovation center. Skolkovo The main activity of the company is research activities in the field of computer vision, data analysis and robotics.

VisionLabs specializes in creating software solutions and services based on computer vision and machine learning technologies. The main product of the company is the recognition platform licionLabs Luna. It allows you to analyze in real time the colossal volumes of photo and video data in order to identify people's faces in them and compare them with multimillion-dollar databases.

Based on this technology, the Face_IS cloud service has been created, which can connect customers with the previous history of their purchases and help build relationships with the brand.

Company decisions for June 2019:

• Template Matching
• Computer vision outside of templates: mapping by key points, recognition of characteristic features, etc., including with the CNN neural network
• Processing large volumes of images from a large number of cameras
• Localize an object in space

Performance indicators

2020

Revenue - more than 1.1 billion rubles

On December 13, 2021, the VisionLabs reported that the company's revenue in 2020 exceeded 1.1 billion rubles.

2x revenue growth

The revenue of AI developer VisionLabs in 2020 increased by 2 times, according to the annual report, the Sberbank ecosystem of which includes this company. The amount of revenue is not shown in the report. But, according to the database of legal entities Contour. Focus"," in 2019 the turnover of VisionLabs amounted to about 391.2 million. rubles Thus, according to estimates, TAdviser according to the results of 2020, it amounted to approximately 782.4 million rubles.

In 2020, the geography of the VisionLabs was expanded to the UAE region. A new direction was also created for thermal imaging software and hardware complexes, the first sales of devices of its own production were realized.

In particular, in the wake of the coronavirus epidemic, the company, based on its technologies, has developed a complex for contactless temperature monitoring and face recognition. According to the developer, the system is able to measure the temperature at the same time in 40 people.

VisionLabs doubled revenue in 2020

The customers of the VisionLabs are large companies in the financial sector and retail, state-owned companies, according to the annual report. An example of retail projects implemented by the company in 2020 is the launch of a contactless face-to-face payment system for the Flash gas station network.

Together with the CST and VisionLabs, a number of projects in the field of "Safe City" were implemented, face and gesture recognition were integrated into the "SberPortal," and the algorithms for face recognition, identification and biometric payment were integrated into the acquiring process and the work of Sberbank ATMs.

And the largest government contract VisionLabs in 2020, according to base "the Contour. Focus", concluded with Nalog Service of FTS: the company delivered to this organization the of the software of VisionLabs LUNA SDK Mobile (system of recognition and the analysis of persons) for the sum about 55.5 million rubles.

Among the key projects in which VisionLabs technologies were involved in 2020, the bank's annual report includes the creation of biometric services for Sber ecosystem companies, including the introduction of a biometric remote access control system (MCUD) from VisionLabs in a number of Russian universities and schools.

2021

Acquisition of 100% of shares VisionLabs MTS structure

On December 13, 2021 Sber reported to TAdviser about sale of 25.07% of shares in the VisionLabs B.V. company of the Intema company, affiliated structure of MTS of AI (LLC MTS of AI, is included into PJSC MTS). VisionLabs specializes in the development of AI solutions in the field of computer vision.

According to the binding documentation, Intema will acquire a 100% stake in VisionLabs in the amount of about 7 billion rubles. The final amount of the transaction will be determined based on the results of meeting the targets of VisionLabs within three years; we are talking about possible additional payments to management and the founders, for Sber the price will not change.

The cost of Sber's share in the VisionLabs since entering 2018 has increased 3.7 times. Sber's total profit from the transaction to income tax will be more than 1 billion rubles, its final amount will be determined at the time of the closure of the transaction.

As a result of the transaction, Intema acquires a controlling stake from the founders of VisionLabs, which before the transaction owned 51.77% of the shares, as well as Sber structures - 25.07% and Sistema VC AFK venture fund - 23.16% of the shares.

It is assumed that after the transaction, the VisionLabs will continue to work autonomously and retain the previous business strategy and team. VisionLabs will remain Chairman of the Board of Directors. Alexander Hanin

"Investment in VisionLabs has brought us an excellent return on invested capital - the value of Sber's stake in the company has increased 3.7 times since entering 2018," said Lev Khasis, First Deputy Chairman of the Board of Sberbank.

According to him, Sber has his own competencies in the field of biometrics, which he develops, including using technological expertise. VisionLabs The partnership with VisionLabs will continue in a number of key areas necessary for the bank to further develop this direction, such as a biometric platform for SmartBio, facial recognition and gestures for SberPortal. In addition, it is planned to continue with VisionLabs the implementation of current contracts and work with ecosystem companies and Sber's partners, Lev Khasis said.

For his part, Vyacheslav Nikolaev, President of MTS, explained that within the framework of MTS AI, the company forms a center of expertise in the field of artificial intelligence, which combines advanced developments, teams and competencies.

"The deal to acquire VisionLabs will strengthen MTS AI's product portfolio in the sought-after direction of video analytics, which we will be able to use both to unlock the potential of the MTS digital services ecosystem and in the interests of our corporate customers in Russia and the international market," the MTS president is convinced.

"Our investment in VisionLabs in 2016 and subsequent withdrawal from this asset is a reference transaction for the corporate venture fund, which brought obvious value to both the fund and the corporation as a whole. On the one hand, at the start of the fund we made the right choice and invested in a business that was successful, significantly increased in value and brought a good return on investment. On the other hand, as a result of the transaction, we strengthened the competence of AFK Sistema Group in such a promising area as artificial intelligence and created the basis for further scaling this area, "concluded Dmitry Filatov, President of Sistema VC.

The parties have already signed legally binding documentation on the transaction. The transaction is expected to close in 2022 after receiving the necessary regulatory approvals.

As of December 2021, the company VisionLabs implemented more than 500 projects in the region computer vision in 37 countries of the world for more than 270 customers from,,, and financial telecommunication transport power retail sectors. More than 1.7 million cameras around the world use ON VisionLabs.

Winner of the competition to protect face recognition systems from hacking

On July 28, 2021, the Skolkovo Foundation informed TAdviser that its resident company VisionLabs won the international competition to protect face recognition systems from hacking Face Anti-spoofing (Presentation Attack Detection) Challenge.

VisionLabs For the third year in a row, the world's best in protecting facial recognition from hacking
Photo source: trashbox.ru

The competition was attended by 195 teams from universities and companies, including Tencent, ByteDance and others.

In 2021, the competition focused on verifying a living person and detecting spoofing attacks (a cyber attack in which a fraudster poses as any reliable source to access important data or information) using 3D masks. As of July 2021, this task is considered one of the most difficult - depending on the quality of the mask, even a person may not always correctly determine it in the photo.

The organizers increased the complexity of the competition by using photos in the dataset with representatives of several races, various 3D masks (transparent, plastic and silicone) and additional attributes, such as glasses or a wig, which hid details. Data on the most difficult attack - with a silicone mask - were opened only 10 days before the end of the competition to exclude manual markup.

Also, one of the conditions was a ban on the use of external data and pre-trained neural networks. This limitation made it possible to choose a winner based on new ideas and optimizing the model architecture, rather than the amount of data for training. Among all the participants, the VisionLabs team created the most effective Liveness algorithm, which with an accuracy of 97% distinguishes a living person from his substitution in a 3D mask.

The results of the competition were summed up by the following metrics^[1]:

• APCER - the percentage of undeclared attacks in which participants' algorithms identified fake as a real person;
• BPCER - percentage of misidentified people where participant algorithms identified real people as fake;
• ACER - APCER and BPCER mean.

Creating liveness algorithms with high accuracy even for complex spoofing attacks such as 3D masks helps make solutions based on computer vision more mass and affordable. We are proud of our research team, which continues to develop this topic from year to year and provides high security for VisionLabs products. Developed counteraction algorithms froda will strengthen our industrial solutions for the financial industry,,, retail where transport face payment and remote verification are especially in demand, said Dmitry Markov, CEO of VisionLabs.

In 2019, Liveness VisionLabs algorithms became the best in recognizing fakes on multimodal data from depth cameras, and in 2020 - determining attacks using printed photos, video from a phone or tablet screen, 3D masks on an RGB frame set.

The Russian school of video analytics is traditionally one of the strongest in the world, which is confirmed by such events. The victory in the international competition VisionLabs once again confirmed its status - one of the most professional developers of video analytics systems. The company's solutions and algorithms are constantly improving, making it possible to combat the actions of intruders and make payment methods for biometric data safer, noted Sergey Khodakov, Director of Operations of the Information Technology Cluster of the Skolkovo Foundation.

2020

Opening of office in Dubai

On November 25, 2020, VisionLabs told TAdviser that as part of expanding its presence in the Middle East, it opened an office in Dubai. It will provide support for presales and technical support for pilot projects in the region, as well as copyright supervision of the company's product implementations.

VisionLabs opened a new office in Dubai. Source photo: old.sk.ru

The office VisionLabs open without attracting third-party investments. This is one of the planned steps in the framework of the development strategy for 2020, which the company was able to implement, despite the economic crisis caused by the coronavirus pandemic. This office is located in the Dubai Internet City Information Technology Park (DIC), created as a free economic zone and focused on creating an ecosystem for global technology companies.

VisionLabs implements large-scale projects in many regions of Russia, the Middle East, Southeast Asia and the USA. The main priority of the company abroad is sales through partners. Biometric products are VisionLabs used by the police of the UAE capital Dubai, in creating a smart city in Singapore, to introduce a recognition system in school buses in the United States.

For VisionLabs, the Middle East is the second region in terms of revenue. The company works not only with the UAE, but also with India, Africa and the Persian Gulf. The most popular here is the use of the Luna biometric platform in the infrastructure of "smart" and "safe" cities, transport and passenger traffic management, KYC solutions for banks. In addition, in a difficult epidemiological situation, contactless payment solutions for face biometrics in retail, as well as remote verification and identification for government and financial institutions, allowing access to important services without leaving home, became especially relevant.

The company's future plans for international development include strengthening its presence in North and Latin America and opening an office in the United States. Another point of development is the promotion of its own devices in foreign markets: the Luna Ace biometric access control and control terminal and the Luna POS biometric payment terminal.

The opening of an office in Dubai is another step towards strengthening the company's position in the international market. This will allow VisionLabs to expand the influence on the entire Middle East region: to more actively develop our expertise there and work effectively with key partners and vendors. In a number of countries, the promotion of solutions is impossible without the creation of an office, since it is perceived as a guarantor of stability, so its appearance is a logical stage in the framework of harmonious growth of the company, said Anton Nazarkin, Director for International Development. VisionLabs

I am glad that the economic crisis caused by the coronavirus pandemic did not interfere with our plans. The open office team will include business development managers and technical specialists, which will strengthen sales support and help VisionLabs play a key role in growing the use of artificial intelligence in the region, said Bushan Kate, regional director of VisionLabs IMEA.

Leader in independent testing of NIST facial recognition algorithms

The company VisionLabs, specializing in the field computer vision and, machine learning on August 25, 2020, announced TAdviser that for the third time in a row it became the first in independent testing (algorithms facial recognition US National Institute of Standards and Technology NIST).

The NIST rating is an authoritative source of efficiency for global recognition systems. Solutions are tested on a multimillion-dollar closed photo database, which allows you to get objective results. Testing is performed on several types of images - from reference photos to documents to pictures from webcams or phones. The main goal is to identify the most versatile facial recognition algorithms that work equally successfully in a variety of application scenarios.

The presented solution VisionLabs became the best in several categories at once:

• Visa - studio photo for documents;
• Border - pictures taken during passport control at the airport;
• Visaborder - comparison of images of two previous categories;
• Mugshot - "police photo," portrait for any file cabinet.

The Russian VisionLabs recognized as the best in facial recognition

In total, more than 140 companies from around the world participated in the testing. The results of the studies showed that the rate of recognition of algorithms is VisionLabs 1.5 times higher than that of all vendors presented in the top of the rating. In addition, by optimizing the architecture of neural networks and efficiently storing biometric templates, the company's solutions require 4 times less memory on the client side, the VisionLabs emphasized. This is especially important when working with large databases of banks, transport and industrial infrastructure enterprises.

The tests also proved that the algorithms VisionLabs work on photographs of any racial groups and can be applied around the world. For example, for a Visa datasheet containing images of people of all ages from more than a hundred countries, recognition accuracy was 99.75% with a one-per-million false positive probability.

Appointment of Dmitry Markov as CEO

2017

Sberbank buys 25.07% of shares

In November 2017, Sberbank announced the signing of an agreement to acquire a stake in VisionLabs. As a result of the transaction, Sberbank will receive a 25.07% stake, the remaining shares of the company belong to the founders of the company (Alexander Hanin, Ivan Laptev and Alexei Nekhaev) and the Sistema VC venture fund .

According to SPARK-Interfax, before the transaction, the founders owned 62.5%. The transaction through which Sberbank intends to provide biometric access to any of its services or services is concluded through the internal structure of a credit institution - the Directorate for the Development of Digital Business, which is engaged in the development of the bank's ecosystem business.

The investment in VisionLabs is the first step of Sberbank to build a biometric platform for Sberbank's ecosystem, which, along with its face, will also use recognition by voice, retina and other biometric factors. After a series of tests and pilots of the VisionLabs LUNA platform, it was decided to build an identification platform on its basis. This will allow you to create a biometric identifier for access to any service or service to Sberbank customers. The biometric platform will be used not only within the Sberbank ecosystem, but will also become available for the market.

Dmitry Dyrmovsky, general director of the Center for Speech Technologies (develops personal recognition systems for voice and face), believes that the cost of the VisionLabs could have grown to 1.8 billion rubles

Sberbank acquires 25.07% of VisionLabs for implementation of facial recognition system

The deal will support the development of VisionLabs projects in international markets, including those in Europe, the USA and Asia. As part of the transaction with Sberbank, the parties reached an agreement to attract a potential international strategic investor to the company's shareholders in the future.

As the founder and general director of the VisionLabs, Alexander Hanin, told Vedomosti, using the company's platform at Sberbank will be the largest biometric project in Russia and one of the largest in the world in terms of the number of customers. Companies will work together not only in financial markets, but also in the areas of retail, insurance, advertising, access control and security, he added.

At the same time, Sberbank's investments will not affect the work of VisionLabs with other banks, partners of the IT company told the publication.

A major IT investor in a conversation with the newspaper said that to service the internal order of Sberbank it would be wiser to buy not a share in the company, but its services. Russian companies are comparable both in technology and in business potential, and the market for IT products for recognizing people will grow, as such systems are in demand in retail, to trade urban infrastructure and security systems.

Perhaps Sberbank wants the VisionLabs to grow on the domestic order, and then the company began to offer services in the market and become the dominant player in the industry, the investor suggests.

Earlier, Sberbank introduced facial recognition technology in an Internet bank for business, it is assumed that more than 256 thousand corporate clients of Sberbank can use the technology (16% of the total number of business users of an Internet bank).^[2]

Partnership with Nvidia

VisionLabs announced in September its participation in the global program Nvidia Metropolis, which aims to unite technology vendors from around the world and apply deep training to create safe and smart cities.

2016

MTS owners invested a third of a billion in Facebook and Google partners

The Sistema Venture Capital venture fund (Sistema VC), created by AFK Sistema, carried out its first transaction in early summer 2016. The fund invested p350 million in Skolkovo resident company VisionLabs, having received a 25% stake for this. The valuation of the company itself amounted to p1.4 billion.

Founders VisionLabs experts in the field of computer vision Alexander Hanin, professor at the Paris National Institute for Research in Computer Science and Automation Ivan Laptev and expert in the field of IT technologies Alexei Nekhaev. The deal with Sistema VC was carried out through an additional issue of VisionLabs shares. Control of the company remains with the founders.

Earlier, VisionLabs attracted subsidies from the Ministry of Education and Science and the Bortnik Foundation (p30 million), and also underwent acceleration at the Internet Initiatives Development Fund (IIDF), which before the deal with Sistema VC owned a share of 2.2%. According to the head of the FRII, Kirill Varlamov, the fund invested p1 million in VisionLabs. The share received by the fund was acquired by Sistema VC^[3]

VisnLabs has created an open computer vision platform along with Facebook and Google

The company VisionLabs, with the support of Facebook and Google, implemented a global open-source project in the field of computer vision. The VisionLabs team integrated two popular developer libraries - OpenCV and Torch.

The project OpenCV bindings for Torch is addressed to the community of developers and researchers. Work on it was carried out with the support of finance and personnel resources from Facebook and Google.

A joint project with Facebook and Google started in 2015. For two IT giants, the topic of deep learning of neural networks and artificial intelligence turned out to be interesting. The most popular developer tools in this area are the open library of computer vision solutions OpenCV and the open library for computing with wide support for Torch machine learning algorithms. Each of them has thousands of users. The company VisionLabs took up the integration of these two libraries.

Google and Facebook funded this work and tested its results.

2014

In 2014, VisionLabs signed a contract with the Equifax Credit History Bureau in Russia to launch a facial recognition system at the interbank level. This system is used as part of the service to counter credit fraud, more than 20 largest banks are connected to it.

Notes