ďOnce We Win a Contract, We Donít Rest Until the Customer is HappyĒ
Interview with Yuri Stoyanov, Chairman of the Board of Directors of Stilsoft Group.
Yuri Stoyanov: born 1972, graduated from the Stavropol Polytechnic in 1994 with a degree in ďElectrical Power Systems, Networks, and Cyber ControlĒ. Began his career at the Stavropol Region state administration, and regional government agencies. Holds 10 patents for inventions in microelectronics and security systems. Head of Stilsoft Group since 2002.
Q: According to some estimates, the global market for face recognition systems is worth about 350m dollars. Do you agree with that figure?
A: Itís very hard to say. I believe itís actually much higher than that. There are, after all, many segments in that market. There is the law-enforcement and security segment, where these systems have a special role to play. There are also biometric access control systems, where the priority is to lower the costs because there is already no lack of expensive and cumbersome systems. We have an R&D program in that segment. And then there is the commercial segment; the 350m-dollar estimate probably refers to the size of the market in that particular segment. There is also market segmentation in terms of the particular services (such as client authorization in online banking, government service bureaus or various online applications).
Any student can put together a simple face recognition system, and it will work fine so long as there are only 100 or 1,000 faces it has to recognize. There are already home entry intercom systems available on the market that recognize the faces of the people living in a condominium; its price is only a little higher than average. The productivity of such a system isnít very high, but thatís not a problem for the people using it. These systems are the future. They will become fairly widespread in another five yearsí time.
Our products offer several key advantages. They work very will with low-quality photo imagery, of the kind usually available to the law-enforcement agencies. A photo may have been taken at night, in low-light conditions, from the bushes, or using a mobile phone, so you can barely see the face. Our product can handle such photos. There are usually certain minimum quality standards that a photograph must meet in order to be usable in face recognition systems. The advantage of our system is that it can work with imagery that doesnít meet those standards. These include e-fit images or photos of dead people; when a person dies, the geometry of their face changes and face recognition systems fail to identify them properly. But our system uses special algorithms that enable them to handle even such difficult cases.
Itís next to impossible to defeat our system by trickery. The behavior of a person who is trying to evade a face recognition system will stand out so much that he or she will be caught for that reason alone. For example, they will walk around in a hat pulled down over their eyes, with a false beard, etc.
Our system is already being used very actively by the Russian law-enforcement and security agencies in nine Russian provinces. It has already helped to improve the percentage of crime cases solved, sometime by a factor of ten or more.
Q: What is the breakdown of your biometric equipment sales by segment?
A: We have next to no sales in the commercial segment. We would like to change that situation, but for now, sales there are barely above zero. There are a few one-off contracts, but the bulk of our sales are in the access control systems and special systems used by the law-enforcement and security agencies. Let me emphasize that the law-enforcement segment isnít made up of ordinary access control systems. These are not just face recognition systems, but systems with important additional capabilities (such as control panels, fingerprinting, face recognition for inmates and prison visitors, etc.). So itís not just an access control system but a system fit for use in a sensitive government facility, including jails. It has more functions and a different setup.
Q: Do you export any of your biometric systems?
A: Of course, and not just to the CIS states but also to countries farther afield. There is a lot of interest in the export markets in our IP and analytical capability. The European and US systems still remain very expensive, whereas Russian companies have always been able to offer a similar level of capability and quality, but with a much lower price tag.
Q: What about the Chinse presence in that market?
A: The Chinese are mainly represented by Baidu. They have some good products Ė but, just like Google or Apple, they use their technologies only in their own products. As a result, they arenít crowding other companies out of that market. For example, you canít buy a Software Development Kit (SDK) from Apple and use it in your own device. You canít use their algorithms in your own system. At the very best, you can use their cloud-based services; in other words, you submit a request to an online service, pay a fee, the service recognizes the face you have submitted and returns the result that you can then use in your own system. But such a setup requires a constant high-quality Internet connection. That is not an acceptable requirement for the law-enforcement and security services, although civilian customers can probably live with that.
Q: So the business model used by the global companies prevents them from crowding out the smaller market players?
A: Exactly Ė and that model is unlikely to change any time soon. I doubt, for example, that they will start offering any products that could be used by the law-enforcement/security agencies. And after all, these agencies would never give a US-based company access to their own databases in the first place. They could offer some limited access for testing purposes, on a small and restricted scale Ė but never comprehensive, nationwide access.
Also, many countries would rather entrust such sensitive projects to Russia than the United States, even though their relations with Washington may actually be better than with Moscow. They believe that Russia is more likely to desist from misusing its access to their databases, and that it wonít use that data for its own purposes Ė whereas the Americans would do that without a shred of doubt.
Q: We are aware that [the Russian arms export intermediary] Rosoboronexport is marketing the Bezopasny Gorod (Safe City) product to foreign customers. Is it one of your own products?
A: It would be wrong to say that Rosoboronexport has placed all its bets on us alone. But we are one of the leading actors that Rosoboronexport regards as potential candidates for that program. After all, our company has high-level capability in video analysis, and Iím not just talking about face recognition but other functionality as well. We are also one of the few Russian makers of high-quality surveillance cameras, including pan-and-tilt models.
We also offer various accessory products and services, such as secure communication channels, traffic enforcement cameras, etc. That is in fact one of our strongest suits. We are the Russian market leader in traffic enforcement cameras. According to various reports, we were the top Russian supplier of such systems in 2017.
To summarize, we are one of the strongest players in that market, and Rosoboronexport regards us as one of the potential partners in marketing such products and services to foreign customers.
We took part in the Safe City promotional activities in Vietnam, where there was a lot of interest in that project. Some high-level negotiations have already taken place. Progress is being made, and I believe that the project can be very promising if Rosoboronexport continues to work with us. We have all the necessary competencies and resources to meet city officials anywhere in the world, tell them about this system, design it from scratch (we have a dedicated design department for that), make that system secure, manufacture the equipment, install it, and launch the entire system. I must admit that we prefer to outsource the installation work to our partners Ė but we rely on our own specialists for launching and fine-tuning our systems. We have close to 100 field engineers who work on site during the launch phase. They also make sure that the customersí own specialists who have been trained at our training center donít make any mistakes. They supervise the entire installation and launch process. We have already polished all the phases of that process.
We also provide reliable technical support and maintenance services. We are a safe bet in that sense because we donít just make promises Ė we already provide these services to our customers.
Q: What is the basic concept of the Safe City product?
A: Safe City consists of several elements. The first is law-enforcement, which boils down to having CCTV cameras in crowded areas. These cameras perform their own video analysis, because no human operator can keep an eye on 1,000 video feeds at the same time. Human operators must only become involved when the system flags up something unusual for their attention.
The second element is incident management. When an emergency happens, the operator must have a list of actions they must complete. That includes integration of various communication tools; in other words, a call must be made Ė preferably in an automatic mode, with elements of AI involved Ė to the appropriate emergency response service.
The third element is the traffic enforcement cameras. Right now, this is usually a separate element because it generates large revenues for regional governments, and therefore has its own dedicated funding. But we regard it as an integral component of the Safe City system. After all, the same cameras can be used to record information about all the cars on that stretch of the road, and to store that information in a common database.
The fourth element is of interest to the Emergencies Ministry. It includes loudspeakers to inform members of the public of any emergency situations, as well as regional and local emergency alert systems. These instruments are increasingly being used not only in emergency situations, but also as a city management tool (for example, to invite members of the public to various events, etc.). The fourth element also includes monitoring the performance of the security and fire safety systems installed at sensitive facilities. For example, a fire alarm should not only detect a fire but alert the fire department automatically, bypassing various intermediate links in the chain. These systems already exist and have been implemented. There are several suppliers in Russia who have rolled them out in various parts of the country.
Q: So, had such a system been installed in Kemerovo [where a deadly shopping mall blaze killed dozens of shoppers in late March], there would have been fewer casualties?
A: Definitely. For example, one of the core features of our system is monitoring the performance of various safety and security subsystems. We monitor the performance of various detectors, such as motion detectors along the perimeter of a military facility or fire detectors at schools. If a detector is malfunctioning or switched off, the system generates an alarm signal after a certain period. It can be set up in such a way that failure of a certain percentage of detectors triggers an automatic alert signal. Our system has been successfully tested by the Federal Penitentiary Service (FSIN), and we plan to install it at all the penitentiary facilities run by FSIN. It is a comprehensive system that includes such functionality as guarding the perimeter and fire detectors; we also plan to add new and fairly unusual functions.
Speaking of the incident management system, it is unique inasmuch as it can include a huge variety of different elements. That includes standard service and maintenance procedures for various rooms and facilities, or a subsystem that monitors the performance of various electronic equipment issued to FSIN personnel. All that information can also be fed to the head office. Of course, the head office personnel in Moscow arenít going to keep an eye on every single detector or fire alarm, but they will keep track of the overall state of the system and its technical performance. Whatís more, all that data will be generated automatically.
Q: So it will be impossible to fool the system or to tinker with it?
A: No. That will be impossible by definition. Even our own software engineers wonít be able to do any unauthorized tinkering. They may be able to disable the alarms for one or two individual detectors, but not on the scale of the entire building or facility Ė that would be completely impossible.
Q: But arenít such electronic systems vulnerable to hacking attempts?
A: That problem can actually be resolved quite easily. All the Russian security agencies are extremely wary of such risks, and very few, if any of them rely on publicly accessible networks. They all build their own isolated networks, which arenít connected in any way to the Internet. For isolated local networks operated by individual government agencies, the likelihood of penetration by some external malicious actor is miniscule.
Our systems also have data redundancy, replication, and emergency recovery mechanisms. These have already been tried and tested.
Q: Are you developing any unmanned vehicles?
A: We are. Our drone is a complex consisting of the unmanned aerial vehicle itself and a container. It is activated by the tripping of one of the perimeter detectors, or when an alert is triggered on the scale of the entire Safe City. The operator presses a key on their computer, the drone leaves the container and flies off on an aerial observation mission. Then the operator presses another key, whereupon the drone flies back to its container and begins to recharge its batteries. We are about to start testing it at one of the MoD units. The Rosatom nuclear corporation and the Russian National Guard have also expressed interest, because the drone incorporates some advanced technologies that no-one else in Russia has mastered.
Our developers have resolved many technological challenges. The drone is well-protected from electronic jamming. It cannot be taken over by jamming systems; it will automatically escape from the jamming ray catchment area and fly back to its home base. It can cope with winds of up to 17 meters per second, it is equipped with an infrared-capable video camera, etc.
The bad news is that the Chinese are making it difficult for us to win a share of the global market, but otherwise everything is fine; all the components we use are Russian-made.
Q: You mean that the Chinese offer similar hardware?
A: Yes, but without the launch container, in which the drone sits in an autonomous mode. The Chinese drones themselves are not much different from our own products, though they may have some advantages in some individual areas. After all, they have a team of several thousand engineers; they have been doing it for several years now, and they have a lot of valuable experience.
The Chinese are clearly ahead of the pack in the mass-market drones segment and in simple military applications. They are even ahead of the Americans and the Israelis. Speaking more generally, the Chinese doctrine of becoming the global leader in high tech by 2025 is working, and it scares me. China is becoming a global leader in an area which has long been one of Russiaís traditional strengths, and I mean high tech.
Q: You have mentioned that you make your own cameras. Does Stilsoft make its own optics and sensors?
A: No, we donít make the optics. We import them from Japan and Taiwan, which offer high-quality products at a reasonable price. Thereís simply no point in us to try making our own. But we make our own optical elements for the thermal imagers. In that segment, imports are still expensive so it makes sense to make our own. We buy the raw materials in Krasnoyarsk and make the lenses in St Petersburg. The final assembly is handled in Stavropol. There are many economic obstacles.
We buy the sensors from Sony, but all the other components and the algorithms are of our own making. Whatís really important here is software; after all, everyone uses the same sensors, but the cameras in which those sensors are used are all different.
Q: To what extent has your business been affected by the sanctions?
A: For us it has long been a matter of principle not to use any suppliers from Germany, the United States, or other countries that could potentially restrict exports to Russia. Many of the chips we use cannot be covered by any of the export restrictions because hundreds of millions of them are made every year, and weíll find a thousand different ways to import them from a huge variety of suppliers. So the Americans do not even bother trying to restrict such exports.
But there are also other chips, the ones used in military or research applications, where the global sales are measured in only a few hundred or a thousand units, so itís relatively easy to track the end user and impose sanctions. We donít import such chips as a matter of principle. We try to find similar products elsewhere, we develop our own, or come up with different technical solutions.
Take, for example, the radio ray detector. Most of the Russian manufacturers use a phased array antenna they import from Germany. That kind of product definitely falls under the scope of sanctions. To shield ourselves from the impact of sanctions, we have developed our own antenna. Yes, it cost us a lot, we have spent a whole year developing it from scratch, but we are now immune to sanctions. We arenít the only Russian company to have done that, but most continue to import antennas rather than make their own.
Q: Are there any Russian companies whose product range is similar to Stilsoftís?
A: I donít believe there are. We make our own hardware - including motherboards - and software. We also make the outer housings. I am aware of only one Russian company that is similar to Stilsoft Ė Iím talking about Eleron, which is based in Moscow. It has been in business for 50 years, it employs 2,000 people, and it is part of the Rosatom corporation. In fact, Eleron supplies security systems to the rest of Rosatom. I donít even regard them as our direct competitor because they are so much larger than ourselves, and they are the sole supplier of all security systems to Rosatom. But in terms of the level of technology, we are definitely ahead of them. They rely on other suppliers for many components; for example, they donít have their own video processing capability, and buy video modules from other suppliers. But their detectors are excellent, as well they should be because Eleron has been making detectors for 50 years.
There are other companies that specialize in such areas as detectors, drones, high-level software or controllers Ė but none does all these things at the same time, as we do. Of course, it took us a while to reach the stage where we are now.
Q: You have been in business since 2002?
A: Correct. At the beginning, it was just myself. Then I hired one other person. Stilsoft was not founded using the assets of any existing Soviet or Russian company, nor did we receive any large investments. I managed to assemble a team that built the entire company from scratch.
Q: Your main customers are law-enforcement and security agencies?
A: Thatís right. They account for 98 per cent of our revenue.
Q: Arenít you worried about such a heavy reliance on government customers? Are you trying to expand into the commercial sector?
A: We are constantly working on that. But the commercial sector alone cannot sustain us, we would have to cut two-thirds of our workforce. It is a much smaller market, and we donít specialize in that area. We offer expensive and uniquely complex technology that nobody else in the world does. I am talking about specialized software, complex systems, etc. Sometimes larger companies win a contract and try to copy our solutions but fail despite being given a lot of time and money. We, on the other hand, already have tried and tested solutions that we can assemble like Lego bricks to make the product our customer requires.
For example, if tomorrow we are awarded a contract to make video monitoring systems for the entire country, we can do that, and very quickly. It would take us months, not years. Other companies, the ones that donít have any experience of building such large systems, wonít be able to do that.
Q: How much interest has the Safe City program attracted from Russian cities?
A: Moscow has a lot of CCTV cameras already installed. There are roughly 200,000 of them, and these are just the ones that feed their signal to the central Situation Center. St Petersburg has almost 100,000. Safe cities cannot do without cameras and security systems. There are not many companies that have the expertise to build such comprehensive solutions. But Moscow and Moscow Region have some fairly capable technological solutions already up and running.
The downside is that everyone and his dog is trying to win a share of that market, because they think itís all very straightforward and the money is good. In actual fact, itís a very complex undertaking. It sometimes happens that a large company wins a contract and thinks it can do it. But the winner proves incompetent; the money is spent but the system doesnít work properly. It happens all the time. The technical specifications can be written in a way that formally, all the requirements are met, but the system doesnít work.
There are only a handful of companies that can actually deliver, and we are one of those. If we win a contract, we donít rest until the customer is satisfied.
Q: Are there any ďsafe citiesĒ in Russia that you think of as a showcase for your technology?
A: In terms of entire cities or towns, there is a town of Mikhaylovsk, not far from Stavropol. The system we have installed there is not large, but it includes all the components that a proper ďsafe cityĒ should have, such as video monitoring, video analysis, face recognition, situation and incident management (with personnel of a private security company monitoring the events), and a license plate recognition system at the town limits that monitors all the incoming and outgoing road traffic. In other words, all the required subsystems are in place. Our task was to equip the town with all the technical solutions required by the mayor. And itís not just a Potemkin village; we have really built a ďsafe MikhaylovskĒ, making all the necessary changes and adjustments along the way. The town now has a complete and comprehensive security system.
Q: Do you have any ďbefore and afterĒ statistics?
A: As soon as the system is installed, the percentage of crime cases solved goes up sharply; it can easily double or even triple. The number of deaths on the roads goes down, because there is less speeding. Whatever some might say, these systems really work and save peopleís lives.
Interviewer: Andrey Frolov
Centre for Analysis of Strategies and Technologies (CAST)