VisionLabs Luna

The main articles are:

• Video analytics (terms, applications, technologies)
• Biometric identification technologies

The VisionLabs LUNA software complex is based on an innovative pattern recognition technology, officially one of the top three in the world in terms of real-world recognition quality. This is confirmed by tests on the independent University of Massachusetts database - Labeled Faces in the Wild.

The solution is designed for processing photo images and streaming video. It includes the functions of face detection and tracking, quality assessment, extraction of key points of the face and coding them in the form of a unique compact key (descriptor). Due to their small size, the resulting descriptors are very effective - they support ultra-fast searches in databases with billions of photos of faces. Thanks to this, VisionLabs technology works up to 200 times faster than similar solutions. VisionLabs is one of the first companies to develop and implement a system for identifying and verifying individuals using convolutional neural networks (CNN) in commercial products.

2024

Use in cyberdogs from the "Laboratory of New Products"

VisionLabs, together with the Laboratory for New Products, presented cyberdogs with integrated intelligent video analytics technologies. The solution was tested at the construction site of the International Medical Cluster Foundation (MMK) and is aimed at improving the level of occupational safety and preventing safety violations. VisionLabs announced this on December 5, 2024. Cyberdog is equipped with an intelligent vision computer system VisionLabs LUNA. Read more here.

Compatibility with Alt OS

VisionLabs and BASEALT entered into an agreement on a technological partnership aimed at developing joint import-independent solutions for the corporate, transport and public sectors. The first step of the collaboration was comprehensive testing and compatibility of VisionLabs' LUNA computer vision platform with the Alt family of operating systems. VisionLabs announced this on January 24, 2024.

The products of the companies are completely Russian developments, they can be used as part of import substitution projects to ensure uninterrupted operation of the technological infrastructure. The solution, which is jointly developed by BASEALT and VisionLabs, will allow you to create and deploy services based on computer vision in a secure OS environment.

BaseALT pays special attention to the compatibility of operating systems with domestic infrastructure products, such as the LUNA computer vision platform from VisionLabs. We strive to provide users with the opportunity to build a digital environment entirely on domestic products, - said Evgenia Krynina, Acting Commercial Director of BASEALT.

Users will have access to a wide range of functionality in the Alt OS environment: Face Pay for the transport sector, biometric services at critical infrastructure facilities, scenarios for commercial biometric systems and much more.

The LUNA computer vision platform allows you to authenticate a person by face, as well as recognize additional attributes: gender, age, emotions, clothing elements, behavior features, and can be used within the framework of smart city solutions. LUNA PLATFORM solutions are already used in banking and government services, in industrial enterprises, in retail and in the telecom sector.

The Alt family of operating systems allows you to create IT infrastructures of any scale. They work on most hardware architectures, including with support for Russian processors (Elbrus, Baikal, ELVIS boards, YADRO IT equipment).

The stable branch Sisyphus"," tested for vulnerability by the BASEALT command, serves as the foundation for building all the company's products. The storage facility is under jurisdiction and within the territory. of the Russian Federation Alt OS is implemented and supported by the company's partners at commercial enterprises public sector and in more than 1 million throughout computers. Russia

{{quote 'VisionLabs and BASEALT are among the leaders in their fields, so the joint solution of companies is especially important in the market. Our partnership will allow customers to effectively use computer vision technologies in systems that fully meet security requirements, including legislative ones, "said Ilya Romanov, Commercial Director of VisionLabs. }}

2023

Astra Linux Special Edition Compatibility

VisionLabs and Astra Group have completed comprehensive product testing as part of a technology partnership with software developers and Ready For Astra equipment manufacturers. VisionLabs announced this on December 11, 2023.

As a result of the tests, the compatibility of the current version of the LUNA computer vision platform with the Astra Linux Special Edition operating system was confirmed. The correctness of the software in a protected OS environment is officially confirmed by a certificate guaranteeing the operability of joint solutions.

computer The LUNA vision platform allows you to verify a to identify person's personality, as well as determine their emotions, gender, age, elements of behavior and attributes. The products used in the clothes product are algorithms fully developed by the VisionLabs team and are recognized as one of the most accurate and fastest in the world. LUNA PLATFORM is used in all areas: smart city bank state services, face payment, multi-camera tracking for building heat maps, etc.

Astra Linux is certified domestic operating system with built-in patented tools information protection for stable and safe operation IT infrastructures of any scale. The OS allows you to process information limited access: from to personal data the mystery state of "special importance."

Both solutions are included in Register of Russian Software"" Ministry of Digital Development of the Russian Federation and meet industry standards. import substitution The compatibility of LUNA PLATFORM with the Astra operating system Linux will allow the use of joint solutions of companies for government services and at sites. critical information infrastructure Another area of ​ ​ cooperation between VisionLabs and Astra Group is the implementation of application scenarios biometrics within commercial biometric systems, including access control and control, authorization in your personal account, etc.

VisionLabs Fast Track Registration Solution

VisionLabs on July 6, 2023 announced the development of a solution for accelerated registration of participants in events. Biometric identification will be an alternative to checking against lists and presenting printed badges, which will reduce registration time and increase security. The solution also allows you to automatically receive event attendance analytics.

The solution is based on the VisionLabs biometric LUNA platform. Read more here.

2022

At the heart of the VisionLabs suite of solutions for medical institutions

On December 26, 2022, VisionLabs presented a set of solutions for medical institutions based on computer vision.

As specified TAdviser in VisionLabs, the complex for medical institutions includes existing solutions that have been finalized taking into account the specifics of the industry. The complex is based on the flagship VisionLabs products - computer the LUNA vision platform intellectual video analysts and a recognition system. transport LUNA CARS Solutions have already been implemented in more than 15 clinics and medical centers throughout. Russia More. here

At the heart of the Face Pass ski lift access solution

VisionLabs and the DataKrat group of companies presented a solution that provides a pass to the ski track according to facial biometrics. Face Pass will be an alternative to the traditional ski pass, as well as paying for travel on the lift using a bank card or phone. This service recognizes a person even in a balaclava and with a ski mask. This was announced by VisionLabs on November 8, 2022. FacePass is based on the Bars payment and bandwidth system from DataKrat and the LUNA PLATFORM computer vision platform from VisionLabs. Read more here.

As part of the solution for contactless diagnosis of respiratory diseases "Neuronics"

On May 16, 2022, the company VisionLabs"" announced that FlexLab integrates the technology company is in a computer vision VisionLabs line of devices - personal medical assistants aimed at contactless diagnosis of respiratory diseases and diagnosis of the condition at health. diabetes mellitus Use face recognition will allow you to automatically confirm the identity of a person and send results to his personal profile.

To start diagnostics, the user needs to confirm his identity. To do this, you just need to look at the device's camera. Integrated it has computer sight distinguishes a person's Luna platform, regardless of the quality of the lighting, the presence of a mask or other additional attributes on the face. More. here

As part of the service for paying with a glance

In the self-service cash desks, Smartix will become available to pay with a glance - a service from Sberbank and VisionLabs based on computer vision. Sberbank announced this on April 26, 2022. Read more here.

As part of AI solutions for biometric access control

VisionLabs and Itrium have presented an intelligent biometric access control solution that meets all import substitution standards. The solution will help complement the functionality of comprehensive security systems for large corporations, which are especially interested in cross-platform and fast technology scalability. This was announced on April 21, 2022 by VisionLabs.

For identification by face biometrics, the system has a built-in computer vision platform Luna from VisionLabs. It provides accurate and fast recognition of a person, regardless of the presence of a mask, glasses, hats and other attributes. The whole process takes a split second. Also, the liveness algorithms from VisionLabs are built into the solution, which create protection against image spoofing. Read more here.

Inclusion in the Unified Register of Russian Programs

More than fifteen VisionLabs products are included in the Unified Register of Russian Programs for Electronic Computers and Databases. The developer announced this on February 22, 2022. Now comprehensive solutions based on computer vision are available for companies that, within the framework of the import substitution program, are guided by Russian-made technologies.

The inclusion of VisionLabs products in the register will allow them to be purchased at a lower price due to VAT exemption. In addition, customers will be able to count on tax deductions and preferential lending conditions provided for in the second package of IT industry support measures and are under development.

The register of domestic software includes the latest versions of the flagship products VisionLabs - LUNA PLATFORM and LUNA SDK. They are based on human-centric technologies of computer vision, which allow you to recognize not only a person, but also his attributes - for example, silhouette, gestures, elements of behavior. The list of Russian programs also includes VisionLabs industry solutions developed in 2021, which allow you to use pattern recognition in access control systems, contactless thermometry and face-to-face payment.

Another product that has been added to the register of domestic software is the LUNA CARS intelligent video analytics system for vehicle recognition. It can define the main attributes of the vehicle - the state number, brand, model, color, type and category of the car, as well as automate event logging, traffic monitoring and traffic management.

All VisionLabs solutions are completely Russian development, as they are based on the company's own algorithms. They are compatible with the products of leading domestic vendors and do not require expensive additional hardware resources on the customer's side due to high optimization of the neural network architecture.

Red OS Compatibility

As a result of the tests, the compatibility of products based on computer vision technologies LUNA Platform and LUNA SDK from VisionLabs with the operating system RED OS from RED SOFT was confirmed. Full compliance with the import substitution policy allows the use of a joint solution in the public sector, including in the subjects of critical information infrastructure. VisionLabs announced this on February 14, 2022.

Also compatible with RED OS is a set of ready-made libraries and neural networks for face recognition and image analysis LUNA SDK from VisionLabs. Due to this, companies using only this operating system will be able to independently develop and add the necessary video analytics functions to their own services and products.

In addition to developing its own products and implementing projects in the interests of, Russian public sectors RED SOFT actively interacts with partners, expanding the RED OS ecosystem. The joint import-independent solution of RED SOFTWARE and VisionLabs, for example, is aimed at ensuring control and security of access using technologies. face recognition Such a decision can be implemented everywhere, including at KII facilities, - commented, Rustam Rustamov Deputy General Director of RED SOFT.

VisionLabs products are based on Russian technologies that meet industry import substitution standards. Therefore, our partnership with RED SOFT is focused on providing the opportunity for strategically important organizations to use computer vision within completely domestic systems. This will not only allow companies to use our ready-made solutions in the RED OS environment, but also develop their own scenarios - from simple authentication to complex services with recognition of silhouettes, gestures, poses and other human-centric attributes, - said Ilya Romanov, commercial director of VisionLabs.

Creating a technology for quick identification of persons for working with multimillion-dollar databases

On February 7, 2022, VisionLabs announced that it had developed a set of additional modules and services of the VisionLabs LUNA computer vision platform, which speeds up the recognition process on large databases while maintaining search accuracy. Especially the increase in the efficiency of the system is noticeable on the bases from 10 million people.

In most cases, the identification process is as follows: the system receives a photo, converts it into a biometric template and compares it with each element in the database to find a match: 10 million persons in the database - 10 million comparisons for one request.

According to the company, most algorithms of other vendors have a linear dependence of the duration of the search on the size of the base - for example, with an increase in the size of the base by 4 times, they are looking for 4 times longer. When using VisionLabs technologies, this relationship is logarithmic. According to NIST testing, searching for a base of 3 million people takes 36 ms, and for a base of 12 million - 43 ms.

This became possible at the expense of software Index. It builds an additional structure with information about the location of descriptors in the database, which allows you to exclude too different faces from the received request from consideration during the search process. This reduces the number of comparisons by several tens of times, significantly reducing. time recognitions

The presented solution also helps to reduce the required amount of computing power, which means to reduce the cost of operation facial recognition system in terms of hardware resources and related costs for their support. ON Index is already used biometric in the platform "," Sberbank which is one of the largest in. To Europe The ability to provide identification or verification by person in a split second improves the quality of service in all client channels. The specialists of the competence center of the division "" helped to test and implement this technology. Biometrics Integration solutions to the existing platform were carried out as soon as possible - within one quarter.

"When standard facial recognition algorithms work with large databases, recognition can take up to a few seconds - this is long enough for users, since services must work in real time. For example, a person wanted to pay for a subway ride in the face. I went to the turnstile, looked into the camera, and he had to wait 2-3 seconds for an answer. During this time, a queue may form behind him, and now imagine such a delay with each passenger. Therefore, our team of researchers and developers optimized this approach and implemented indexed search technology that allows you to identify a person in a split second even with a multi-million dollar base, "- told Sergey Milyaev, head of research projects.

Integration with BioSmart Quasar Access Control Terminal

On February 2, 2022, the company Biosmart announced that access control terminals BioSmart Quasar are now supported identification by person by technology, VisionLabs one of the world's manufacturers in the region. computer vision Accuracy identifications is achieved regardless of the level of illumination of the room of additional attributes - masks, glasses and hats. The joint solution of companies meets the current industry requirements for. to import substitution More. here

2021

As part of the solution for navigating bank customers

On December 17, 2021, VisionLabs announced that they had created a client smart navigation solution with Q-Systems. Smart navigation is based on the VisionLabs Luna video analytics platform, the Orchestra visitor flow control system and integration with the bank's CRM system. Read more here.

Introduction to Finservice platform

The Financial Services Platform of the Moscow Exchange will introduce the VisionLabs Luna Platform facial recognition system into the customer identification process. The use of data validation technologies will improve the quality and speed of service to platform customers. This was announced on November 16, 2021 by VisionLabs. Read more here.

Plan to add VisionLabs facial recognition technology to the BIMData ecosystem

On August 6, 2021, the company, VisionLabs one of the representatives in the field computer vision and, and machine learning the digital ecosystem for constructions BIMDATA announced the start of a technological partnership. The companies have joined forces to create a solution that allows you to increase control over the implementation of construction projects, personnel and the dynamics of work. More. here

Gaskar Group Development Implementation

On March 4, 2021, one of the IT integrators in the field of construction and real estate GASKAR GROUP announced the introduction of vision technology from VisionLabs in its developments. Read more here.

Integration with InOne IoT Platform

On January 27, 2021, the Group of company Systematica announced that the InOne integration IoT platform developed by HeadPoint was adapted to work in geographically distributed structures.

To expand the monitoring capabilities, InOne has added the functionality of LUNA PLATFORM from VisionLabs - face recognition and analysis using neural networks. Read more here.

Integration with Examus

VisionLabs and Electronic Platforms, residents of the Skolkovo Foundation Information Technology Cluster, have developed a product based on LUNA Exam computer vision. It includes the Examus monitoring system and biometric verification when taking an exam or testing based on VisionLabs facial recognition and analysis algorithms. This was announced on January 25, 2021 by the Skolkovo Foundation. Read more here.

2020

Obtaining a certificate of the Ministry of Internal affairs

The biometric platform LUNA VisionLabs received a certificate from the Ministry of Internal Affairs, which allows the company's technology to be used in transport security projects - at railway stations, sea and river ports, bus stations and airports. The developer announced this on October 8, 2020.

Tests for product certification were carried out by specialists of the department. The received document confirms the compliance of LUNA Platform with the Decree of the Government of the Russian Federation No. 969 in terms of requirements for the functional properties of technical means of video surveillance in transport. They relate to such items as providing video verification, visual control, programming of operating modes, determining significant parameters of detected situations (nature, place, direction of movement of objects), etc.

For a modern city, transport infrastructure is an important aspect of development. Increasing the load on interchange nodes undoubtedly requires a corresponding increase in the level of security. Computer vision-based products enable monitoring of the situation and the use of preventive protection measures, which is especially important for crowded locations.Within the framework of the "Safe City" LUNA Platform for October 2020 is already part of the multi-platform solution of urban video surveillance in Moscow. The system is used to improve the safety of residents, control over the content of urban areas and has shown its effectiveness, - said Dmitry Markov, CEO of VisionLabs.

VisionLabs' LUNA computer vision and video analytics platform can verify and identify a person's identity, as well as determine their emotions, gender, age, behavioral elements and clothing attributes. The use of LUNA Platform is possible in all areas of the "smart" city, it is used in projects on public transport, socially significant and industrial facilities, in banking and state institutions, retail.

Russian facial recognition technology recognized as the best in the world for the second year in a row

VisionLabs (part of the ecosystem) Sberbank- a company in the field of systems creation computer vision and - machine learning won first place at the international ChaLearn Face Anti-spoofing Attack Detection Challenge at the CVPR 2020 conference - the main annual to computer vision event. The company won for the second year in a row. In total, there were more than 340 participants in the competition - companies and scientific teams from different countries, including such as Intel Hisign and others. This was announced on March 13, 2020 by Sberbank.

The VisionLabs team presented Liveness technology at the competition, which doubled the results of its closest competitor. The main task of the competition was to verify that the face of a real person is recognized in front of the camera, and not a fake in the form of a printed photo, video or mannequin. In 2020, the organizers complicated the conditions for the competition compared to the past: the competition was held to identify Liveness attacks on three databases of photographs with representatives of different races, training was held on the same base, and tested on the other two. Participant algorithms were evaluated on attacks that were not in training and which are critical in various areas of application of facial recognition, primarily in the financial field. The algorithms of the participants needed to determine whether a living person was in front of the camera or an image from a phone, tablet, 3D mask.

The results of the competition were summed up according to the following metrics:

• APCER - the percentage of undetected attacks, where the algorithms of the participants correctly identified the attack as a real person;
• BPCER - the percentage of incorrectly identified people, where the algorithms of participants determined on real people that this is a fake;
• ACER is the average APCER and BPCER score by which participant algorithms were evaluated.

According to the results of the competition, VisionLabs took first place, twice ahead of other participants.

We are pleased to confirm our high status in computer vision at the international CVPR competition. In 2020, a number of corporations with really strong developments in the field of computer vision and machine learning took part in the competition. Therefore, the victory is doubly pleasant, and we at VisionLabs are very proud that dear experts for the second year in a row recognize our product as the most accurate and effective. We constantly improve the quality of existing products and create new ones based on our own algorithms. The application of our boxed products in various areas of business ensures the most efficient and secure business processes, increased customer loyalty and the quality of the services provided, noted Alexander Hanin, Co-Founder, Chairman of the Board of Directors of VisionLabs

2019

Integration with Yandex Signal Q1

On August 5, 2019, the company VisionLabs announced that algorithms VisionLabs facial analysis was part of a driver fatigue and attention monitoring system that it intends to launch Yandex.Taxi after pilot testing on several thousand. cars The system is able to assess the condition of the driver, after which, if necessary, he receives a warning signal or restriction access on service orders. Neuronet Yandex learned to recognize driver fatigue, and VisionLabs technology helped to do so on the move. More. here

Russian facial recognition technology recognized as the best in the world

VisionLabs, (part of the Sberbank ecosystem), a company in the field of creating computer vision and machine learning systems, for the second time in a row won first place in testing the identity of persons of the US National Institute of Standards and Technology (NIST). TAdviser was informed about this by representatives of VisionLabs on July 4, 2019.

VisionLabs technology took first place in the Mugshot category and entered the top 3 Visa categories. The speed of recognition of the VisionLabs algorithm is twice as fast as that of other participants with similar solutions.

In the Mugshot category (a photo portrait of a criminal where lighting and background are variable and image quality can be poor): facial recognition is tested on a base containing more than 1 million photos of people. The peculiarity and complexity of the base is that it contains photographs of the same person with a significant age difference. The VisionLabs algorithm recognizes 99.6% correctly, with a 0.001% erroneous response rate.

Separately, in the Mugshot category, he passed a test in which the task is to recognize people from photos taken 14 years apart. In this test, VisionLabs took first place (99.5%, with a share of erroneous positives of 0.001%) with the most age-resistant facial recognition algorithm.

In the Visa category, recognition occurs in studio photos in good lighting on a white background. The recognition base is several hundred thousand photographs of people. The main difficulty of the base is that it represents people from more than a hundred countries. In testing, the VisionLabs algorithm correctly recognized 99.5% with a failure rate of 0.0001%, which was the second result among all manufacturers.

We at VisionLabs are very proud that our algorithm is once again recognized as the most accurate and efficient. We are trying to constantly improve our developments to provide the highest quality of technology to customers. With our solutions, customers in various areas of business provide the fastest and safest work processes, increasing the loyalty of their customers and the quality of the services provided,

The table shows the most efficient 1:1 algorithms measured by mismatch rate (FNMR) in several different datasets.

In April 2019, VisionLabs took first places in the Mugshot categories, and also entered the top three in the Visa slice. VisionLabs algorithms showed a good level of accuracy, as well as speeds 2.5 times faster than all major competitors.

NIST began evaluating facial recognition technologies in February 2017. FRVT 1:1 testing corresponds to the scenario of confirming a person's identity by verification by photo. The study is also designed to help the Ministry trade USA identify the world's best suppliers of such software solutions.

In March 2019, VisionLabs won first place at the largest international ChaLearn Face Anti-spoofing Attack Detection Challenge from CVPR 2019, the premier annual computer vision event. The VisionLabs team presented Liveness technology at the competition, which exceeded the results of the runner-up by 1.5 times. In total, 25 teams from different countries took part in the final stage.

Announcement of ID.Abonent using ABBYY and VisionLabs technologies

On June 20, 2019, the company ABBYY announced a company developed GoldenSIM in partnership with ABBYY and VisionLabs mobile application ID.Abonent that allows remote registration SIM cards on the network. telecom operators The technology ABBYY solution and VisionLabs on the fly recognizes, passport data confirms the identity of the user and sends information about the new client to the operator. After a few minutes, the new number is registered on the network, and you can already use it. More. here

Integration with timebook workforce management

On February 21, 2019 Skolkovo Foundation , it was reported that the residents Fund IT Cluster are companies timebook VisionLabs and have entered into a strategic agreement. As a result of face recognition the partnership, VisionLabs' LUNA platform will complement the flagship product timebook aimed at comprehensive Human Resources Management retailers. More. here

2018: Working as a plugin on the Nvidia Drive IX platform

On October 15, 2018, Visionlabs announced the introduction of face recognition inside and outside the car into the Nvidia Drive platform. According to the company, this technology will replace the keys, provide "smart" and safe access to the car, personalization, and control over the driver's condition. Read more here.

2017

Audit of personal data processing process

VisionLabs conducted a survey of its software complex at NPO Echelon in order to determine the list of personal data processed. The examination showed that VisionLabs systems do not provide for the processing of personal data of subjects.

The LUNA software complex, the LUNA Cloud cloud service, as well as a set of LUNA SDK libraries, FaceStream software, and PhotoMaker libraries were examined.

The interaction of software levels and the transfer of VisionLabs information flows involves 5 stages, none of which process personal data. First, the image or video stream is pre-processed and enters the middle-end system in the form of a transformed image of a person's face. The middle-end system sends the image along with the request to extract the descriptor to the back-end server using the LUNA SDK. The back-end system then processes the resulting image, retrieves the handle, and forwards the handle ID back to the middle-end system. The middle-end system then forwards the descriptor ID along with the search request to the back-end server, where it finds the descriptor by its ID and maps it to the descriptors stored in the LUNA database. Then it sends the middle-end identifiers of the most similar descriptors and the degree of their similarity to the system.

In fact, according to the developer, VisionLabs technologies do not imply physical storage of a photo, photos are needed only at the first stage. Then it is converted into a descriptor (numerical value), and further work is carried out with it. The original image cannot be recovered from the handle.

Collecting age-and-sex data for marketing analytics in offline retail

Also in the summer of 2017, the company launched a solution that allows collecting age-sex data for marketing analytics in offline retail. The system allows you to give a probabilistic estimate of the gender and age of customers in the video stream. Such data are processed anonymously, without the participation of personal data. After analysis, this data is returned as a numeric value by the system. They cannot be attributed to personal data, since they do not allow unambiguously identifying a specific person.

VisionLabs and Smart Engines unveil joint KYC class solution

VisionLabs and Smart Engines, Russian developers of image processing and recognition technologies, on July 7, 2017 presented a new Know your customer (KYC) class solution designed to verify new banking customers who came through digital channels without visiting the office.

In general, the novelty is an SDK for implementation on mobile devices and PCs. Face recognition is based on the VisionLabs LUNA platform, which allows you to identify and verify faces in real time. For document recognition, Smart IDReader technology from Smart Engines is used, which implements the ability to recognize the passport of the Russian Federation not only in images, but also in the video stream on mobile phones in on-device mode.

From an architectural point of view, the technology is a set of software libraries for embedding in the customer's IT infrastructure and supports Linux, Windows, iOS, Android. Functionally, the solution provides remote customer service through a mobile application and personal account in a browser.

The joint solution of VisionLabs and Smart Engines will allow you to recognize text data of documents on mobile devices and in the browser and compare photos on documents with "selfies" of customers. In turn, this will help banks reduce their risks with remote banking services when opening accounts and making decisions on issuing loans.

According to the developers, VisionLabs algorithms are resistant to age-related changes, beards, glasses, bright makeup, show high recognition reliability even in low light and work with the simplest equipment (cameras from 0.3 megapixels), allow you to recognize both black and white and color images with equal high accuracy.

The novelty is focused on use in banks, microfinance organizations, insurance companies and other projects that require remote verification of customers before providing services. The technology can also be used in simcoms - devices for issuing SIM-cards and servicing subscribers of cellular operators. At the moment, the simcomat can work only with the participation of an employee of the call center, who remotely compares a person's photo with a photo in his passport.

"Speaking about the recognition of photo images on documents, I want to say that the verification process is complicated, first of all, by the presence of holographic protection on most photo images, which, when scanning a document, can distort data for recognition. The technology that we created together with the Smart Engines partner will allow you to recognize such images and is a comprehensive solution that does not require the purchase of additional equipment or complex integration, "says VisionLabs CEO Alexander Hanin.

From a market point of view, the product has no geographical restrictions on its use. To work on the Russian market, the technology allows processing a passport of a citizen of the Russian Federation, all valid types of driver's licenses, a foreign passport, as well as SNILS, STS and bank cards. The system also supports a wide class of national documents, USA,, Great Britain,, Germany Italy Spain Israel,,,,, China India Japan Kazakhstan and other countries of the world.

The product distribution model is the sale of a license, the cost of which is determined by the number of checks.

Principle of operation

To verify a client in his personal account in an Internet bank, he needs to send a "selfie" to the server and a scanned or photographed document. In the uploaded image of the document, the system recognizes text data that is automatically entered into the application/questionnaire, finds the client's photo and compares it with the "selfie" uploaded by the user.

In the mobile application, the client is invited to take a picture of himself and "show" the document to the system, and she recognizes it in the video stream. At the same time, the recognition process, including the operation of inter-frame integration algorithms, is performed on the device itself in on-device mode, and "selfies," photos from the document and a text file with the extracted data are sent to the customer's server. Comparison of faces in images is carried out on the server.

The bank can also check the client's photo image against its own or interbank database through the credit history bureau for the compliance of the client's data and the photo that is in the passport. Comparison of faces in images is carried out already on the server.

2015

In 2015, VisionLabs and SAP Labs CIS announced the integration of VisionLabs technology with the SAP HANA platform.

See also

• National Biometric Platform (NBP)
• Biometric identification technologies
• Biometrics and Border Control
• IDensic
• Sonda RTI
• Cognitive Passport
• Biometric identification