How can an IT company end up on the sanctions list

In mid-December last year, the European Union adopted the third package of sanctions against legal entities and individuals from Belarus, which includes the IT company Synesis.

In the document, the European Union claims that the company has provided the Belarusian authorities with a platform that can recognize the faces of people who participated in the protests.

The co-founder of Synesis, Alexander Shatrov, told KV.by about how the imposition of sanctions affects the work of the company, how Kipod actually works, and how it all began.

Imposition of sanctions cisco voice jobs

The company does not submit itself to the list of sanctions, as for participation in the competition "company of the year". We did not fight for this "honorary" title, and personally I do not know the reasons why we got there. But I was much more struck by the fact that none of those authorized for making such responsible decisions is immersed in information, especially the part that is absolutely public: what kind of company, who has its founders, who is the director. It got to the point of ridiculousness: in the official magazine of the European Union I was listed as CEO, even though I am not.

In fact, there is a whole European Parliament of the European Union, with investigators and experts who are as immersed in the problem as possible and study everything in detail. If they could not even figure out who the CEO of the company was, where it was only necessary to open a register with public information and read who the director is in the company, the question arises: how deep did they plunge into more complex charges?

And about the reasons for falling under the sanctions - I have no idea. There are no direct or even indirect reasons why we had to be there. But we are there. My opinion? Someone purposefully made an effort to get us there.

Now we are preparing an appeal to the European Union [at the time of preparation of the material, the appeal has already been filed.

Company damage

Now about the direct harm from everything that happens. Instead of engaging in projects and work, we are engaged in absolutely non-constructive and unproductive activities to challenge the decision of the European Union. This is an unpleasant activity that takes time, money and effort.

Surprisingly, people manage to make decisions quickly, but the procedure for challenging these decisions takes much longer. And this goes against the values ​​declared by the European Union. On the one hand, they, for example, accuse someone, include in the sanctions list, which clearly has consequences, but at the same time they do not even give an opportunity to answer and do not ask questions, they immediately say: "Look, you are accused of this."

We are not very worried about mentioning the company somewhere, and it is obvious that the issue is not in reputation losses, but in exclusively economic ones. This applies not even to many partners, but to work with a number of services. We have, for example, Google Suite, a whole group of companies have been working with it for many years. Documents, mailboxes are stored there. Google's European headquarters is located in Ireland, and this is the UK, which traditionally adopts European sanctions. There are sanctions, which means that any economic relations with us are prohibited for European and British companies.

At the same time, “economic relations” does not mean that they pay us. There are also situations where we pay them. And, moreover, as I understand it, these relations are criminalized.

Kipod

Kipod started in 2006-2007. We have been involved in various image processing projects. And at some point we decided to make our own products and hardware solution. Back then, most cameras were analog, and we made a special device to which such cameras can be connected. Analytics was supposed to work on this device (even before the era of neural networks). Then there were also face recognition mechanisms, but they were just not as accurate as they are today.

We realized that making our own hardware and fighting heroically with Chinese companies for the price and quality of production is pointless. Also, we encountered such funny things: if you need to send a test sample somewhere, it will not be easy (you have to go through a huge number of customs-related procedures). So this has evolved into a pure software solution.

Face recognition technology is now one of the basic technologies in the world. And this is not a fantasy or a unique solution available to several companies. The American NIST (National Institute of Standards and Technology) ranks various algorithms that deal with face recognition. A huge number of companies send applications, only 150 are included in the rating. There is no single rating, there are different algorithms and tasks: street cameras with poor lighting and bad angles, or, for example, face recognition like in American films - if you have a face of good quality (visas or passport). There is no such concept: “we recognize any faces”, all algorithms have their own narrow specialization. And companies that make it into the top ten are fighting for hundredths of a percent of accuracy after 99%.

If you want, type in the search engine "facial recognition download free", you can even specify on which platform it should work. And you will find a sufficient number of ready-made and trained neural networks. They will not have 99+ percent accuracy, but sufficient to recognize faces.

If you need to process images from hundreds of thousands of cameras centrally and in real time, you are faced with a difficult task from an engineering and economic point of view. When you have a house or apartment and there are three or even five cameras, then the cost does not play such a big role. The whole difference will be within $ 20-50. When it comes to centralized processing of a huge number of cameras, you cannot solve the problem at any cost - it turns into millions of dollars. Therefore, it is necessary to optimize the algorithms, while increasing their accuracy.

This is what Kipod does. And not only by video analytics, we can also have sound as an initial signal: broken glass, scream, shot. The main thing is that you have a huge number of sensors with "heavy" data for processing, which must be processed in real time, and also allow you to search in the archive.