Where faces are recognized

    In less than a year, VisionLabs , a company that specializes in face recognition technology, has become one of the winners of the Web Ready contest ( GoTech since 2016 ), has been included in the list of 12 finalists of the Challenge UP! Program, organized by Intel, Cisco and Deutsche Telekom AG, as well as attracted 350 million rubles of investments of the venture capital fund Sistema Venture Capital.

    Alexander Khanin, CEO of the company, talks about how face recognition technologies are being introduced in the commercial sector, what the future holds for them and how those who fall into the lens treat them, and also identifies solutions that can compete with them.

    Everyone is used to the fact that face recognition technology is a purely state history aimed at ensuring security. But today, technology has also been actively introduced in the commercial sector - every year its share in the overall face recognition market is growing by 30%. Here the effect of the technology is noticeable every day - dozens of scammers who are ready to, for example, withdraw money from banks using fake passports. Savings in the average bank is about $ 100 thousand per month. In the field of security, the opposite is true, the issue of return on investment and return on investment is not paramount; the goals of introducing such technologies do not have to have a commercial effect.

    Banks, retail and car dealerships


    One of the main motives of the business is the relative cheapness of such technology and the benefit compared to other ways of working on customer loyalty and safety. Among the areas where face recognition is becoming popular, it is worth highlighting the banking sector and retail.

    Banks are now number one in terms of technology consumption in monetary terms. For example, last year, according to the National Bureau of Credit Histories, 596 thousand loans with signs of fraud were recorded in banks. Face recognition just helps in the fight against these cases, and also accelerates customer service. For example, technology is being introduced to confirm payments. Classic solutions with SMS confirmation are more expensive. Each sms to the client’s phone with a password for confirmation costs the bank several tens of kopecks, while photography and data exchange are several times cheaper. Often, face recognition is complemented by a fingerprint, which gives maximum accuracy and the inability to fake a person’s data.

    In addition to banks, face recognition actively penetrates into retail. There are also several tasks. The first is the fight against serial thieves. It is necessary to recognize among visitors those who have already come across a theft before. The second task is a qualitative increase in the effectiveness of the loyalty system. A huge number of people come to the store, but few of the employees remember who they are. Face recognition technology provides a tool that digitizes the incoming client stream, understand who and how often returns to them, and select vip clients. Based on all these data, loyalty programs and unique offers for visitors are formed, the workload of consultants and consultants is analyzed, which allows you to create the most efficient schedule for them, and ultimately, the technology makes it possible to create that product range,

    Recently, car dealership owners have begun to pay attention to technology. Of course, mainly representatives of the premium segment. What is the benefit? Usually, when a client hands over a car for maintenance, this is accompanied by a rather long and often unnecessary dialogue with an auto center employee. What happens with the introduction of technology? As soon as the client enters the car dealership, the manager who serves him receives all the necessary data - the make and model of the car, the recording time for maintenance, etc. The number of questions is reduced, and the car owner feels at home, he believes that he is known in person. The car dealership receives a serious tool in the fight for customer loyalty, and, as we understand it, in the conditions of a falling car market, many car centers rely on the service.

    Big Brother


    If you imagine that I will photograph a person in Europe and say what his name is, he will be very dissatisfied with this fact and will most likely go to court. In Russia, everything is different. Yes, the older generation is conservative, but those who are under 30 and a little taller are quite loyal. I think the reason is as follows. Everyone is already accustomed to selfies, technologies of the social network Facebook, which gives a lot of information from personal photos, so the appearance of a new “chip” with face recognition causes a positive reaction. For example: "what a fashionable bank, how cool that I am served here."

    Recently, in one of the CIS countries, we conducted an experiment on customer service at a bank without a passport. 99% of visitors were ready to recognize themselves by their faces. This was a revelation for them, people were glad that in fact they did not need to provide any documents. They are required only de jure.

    By the way, for such a positive attitude, it is worth expressing gratitude to large IT corporations. Social networks and popular mobile applications have helped smooth out the attitude towards the development of our technology.

    What is the future and who is the competitor?


    With the development of face recognition solutions, the privacy issue of processed data is becoming increasingly acute. With those who implement the technology, we agree that we will never allow a person to be named. For example, in retail, a customer is recognized in an anonymous form, the store does not care what his name is, the task is different - to determine what to offer him at the moment. This is the concept of working with the commercial sector.

    But the problem itself still exists. The legislative framework in this regard is very seriously behind the development of technology. I think that over the next 3-5 years a lot will depend on whether they decide in Russia and other countries the issue of giving legal recognition results.

    In the meantime, technology is moving towards expanding the range of possible working conditions. If earlier it was possible to recognize faces only frontally and plus or minus a slight turn, now the angle has increased significantly. The client no longer needs to strain and look directly into the camera, which, of course, affects his comfort. The next task that should be solved is, with the preservation of quality, to expand the volume of databases with which you can compare the results. Today, millions of individuals can be recognized in the databases without errors, but by the end of this year we want to bring this figure to 10 million or more.

    Where will the technology go next? I see several new areas of its implementation.

    The first is the mobile payment segment and biometric acquiring. Payment will be made without a credit card, when the transaction is confirmed using some biometric factor.

    The second segment is robotics and automation. For example, automatic payment terminals in retail, where a cashier is not needed. In robotics, today it is believed that a robot that cannot recognize faces is already a defective robot. For example, he should distinguish who approached him - the owner, who has more authority, or just a visitor to an exhibition. By the way, at the exhibition itself, he must determine whether this or that person approached him earlier.

    The third area is the Internet of things. Any sensor can be intelligent. For example, you open the refrigerator at 3 o’clock in the morning, and he recognizes your face and says that he will not open, since you yourself set such a timing for him.

    The last question is competitors. Two technologies can be distinguished here. In some narrow cases, the identification of a venous palm pattern may compete with face recognition. This is a more or less non-contact method, although it works in limited conditions. Where? For example, an ATM. In some cases, this will be even more beneficial than recognizing the client by face. But while the degree of technology penetration is low, and databases (palm scans) are negligible compared to photographs.

    The second competitor is the phone ID. For example, a GSM signal can be used to determine the fact of a person being in a store. Such a solution will work more accurately than face recognition. But there is a problem. And this is legislation and the concept of secrecy of communication. It is necessary to collect stable bundles of information (who owns the phone), which at this stage is quite problematic. Technologically, this can be done, but not yet by law.

    Where is the border?


    In general, if we talk about the degree of development of recognition technologies, then every year scientists report that the quality is getting better and the number of errors is decreasing, although they are still quite large. From a technological point of view, it is already clear what remains to be done. I think that in the next 1.5-2 years the face recognition problem will be completely solved. Therefore, we look not so much at the development of technology, but at how it is necessary to solve the problem. There is a struggle for the speed of market capture, for the volume of databases and for working business models. And, of course, there is no way to go without the development of legislation. Now it lags behind the level of technology for many years. And not only in Russia.

    Also popular now: