People think that to create a successful startup, all you have to do is to come up with an idea, implement it, and put it on the market. Once it’s done, all that's left is to develop the business. Nevertheless, novice entrepreneurs have to work a lot after getting their very first orders, and it’s the kind of work that isn’t spoken about often: communication with clients, search for distribution channels, urgent adjustment of the product.

Specvideoanalytica, a St. Petersburg company that got a grant from the Innovation Promotion Foundation and became a Skolkovo resident this year, faced all of that. So their path to success was not quite simple.

Skolkovo. Credit: sk.ru
Skolkovo. Credit: sk.ru

Smart barrier

The team’s story began in 2017, when staff members of ITMO’s Machine Learning Lab founded a company together with their colleagues working in the field of industrial automation systems development and introduction. The new project focused on the development of smart systems for controlling the access of cars to residential areas. 

The new startup offered an alternative to pass cards or contacting the security service via an intercom. A computer vision technology allowed their equipment to identify cars that have access and deny access to unregistered cars. What’s more, the system also identified emergency services' vehicles and granted them immediate access, saving precious time.

The startup got the status of a small innovative enterprise with the university having a share in it. The team participated in St.Petersburg International Innovation Forum 2018, MY.TECH 2018, 16th Forum of Small and Medium Enterprises of St. Petersburg, StartUp Ferma and others. But that was just the beginning of their path: the team was to find their main clients, partners, and scale the production. 

St.Petersburg International Innovation Forum 2018
St.Petersburg International Innovation Forum 2018

What’s next

The project offered a hi-tech solution to the acute problem of car-filled yards, but few homeowner associations and managing companies were ready to install the new technology right away.

“At first, only a small percentage of the market believed in our technology,” remembers Natalya Hanzhina, one of the company’s founders. “This is understandable, taking into account the contemporary level of development and trust in computer vision technologies. But there were people who agreed to have our equipment installed. We used their sites to test out technology. What I am 100% proud of is our first stable runs. When someone asks you whether you already have clients, and you can say yes, plus you have a stable working solution ― that’s really cool, it’s like seeing your baby walk for the first time.”

Gradually, the company started to also develop another adjacent market. Thing is, you need to control the access of cars to not just residential areas but also to storage facilities, plants and so on. The owners of some of such locations also decided to give the smart system a try.

Nikita Kulikov
Nikita Kulikov

“This called for making adjustments to the business model,” remembers Nikita Kulikov, the project’s head. “We got partners and also the opportunity to work on different types of objects. Our partners focus on sales and installation of complex security systems, and we’ve focused on vehicle pass systems. In order to do that, we had to make our product look good, like something you’re ready to sell. Before, we supplied elements of our equipment without much regard to the way they look, but from now on, we have a finalized, ready-to-go solution.”

Working on weak spots

Gradually, the company succeeded in moving beyond St. Petersburg and finding clients in Moscow. But at first it didn't go as well as they hopedю According to Nikita Kulikov, their work with their first client in Moscow became one of the hardest stages for their team in the last two years.

“At first, everything worked fine, but there were malfunctions. For example, when the connection with the server let us down. Thing is, at that point, we only installed cameras at the site, and all processing took place on our server. But when we lost the connection, the whole system stopped working. This caused very unpleasant situations and we received negative feedback from the customer. Thankfully, the customer showed understanding, we overcame this difficult stage and continue to collaborate,” he comments.

Vehicle pass system equipment. Image from the project heads’ archive.
Vehicle pass system equipment. Image from the project heads’ archive.

The situation with the cloud solution issues led to the company seeking to adjust their product, offering the client a choice between a cloud solution and a new option where images were processed at their site.

“We did it by optimization of our neural network models,” remembers Natalya Hanzhina. “Technologies constantly progress, and now major companies that produce processors actively participate in the development of deep learning frameworks for accelerating the work of neural networks. For one, Intel developed the OpenVINO framework for the acceleration of neural networks for their own central processors. This is not the quickest technology, but it’s ok for the purposes of processing images and videos.”

The team also improved the algorithm responsible for the automated granting of access to special vehicles so that it would function at new sites.

ICV controller. Image from the project heads’ archive.
ICV controller. Image from the project heads’ archive.

It’s all about the team 

Much as any young team, Specvideoanalytika not only had to search for clients and partners, solve the early problems of their product, but also attract new team members and keep the old ones, which is not always easy.

“Surely, an important factor was that we didn’t succeed in becoming self-sufficient,” says Nikita Kulikov. “There was a time when we had marketing specialists in our startup, had more people working on neural networks and recognition systems. At the peak, we had eight members; now some of them have completed their tasks and moved on, and we continue.”

Natalya Hanzhina
Natalya Hanzhina

“What’s most important for me is that we’ve retained the core of our team”, adds Natalya Hanzhina. “All members of our team ― four co-founders, frontend developer and my colleague, neural network developer, ― have been together since the start of the project, and that’s very important. What’s more, we are still full of enthusiasm.”

Road to Skolkovo

When a project already has a product and first sales, but does not have a stable share on the market and sufficient turnover, grants and other support are especially important. About a year ago, the project got financing thanks to a Start grant from the Innovation Promotion Foundation.

Becoming a resident at Skolkovo was a long-standing goal of the team, and it took two years to reach it. Today, it’s most relevant, as in order to apply for the second round of the Start grant, they need co-funding that can be found at Skolkovo.

“This year, in the times of isolation, we began working on our application once again, read it through and understood that we never included information on some of our important achievements, like winning a grant. Once we did it, they took us in. It seems that the members of the selection committee understood that we have proof of our viability,” says Natalya Hanzhina. “Why do we need it? Well, it’s obviously prestigious, but most importantly, it gives us access to investors. Skolkovo has a pool of private investors working with it.”

The company members expect that the resources that Skolkovo offers will help them make the next step towards self-sufficiency and profit.

Credit: Sk.ru
Credit: Sk.ru

Lots to be done

The project’s current key task lies in attracting new clients and searching for partners. The company also plans to continue developing their project, the priority being improving the recognition quality at acute viewing angles and in bad lighting conditions.

Another service that Specvideoanalytika plans to offer is a remote online control center, that will help small homeowner associations save on having a concierge. 

“An online operator is a person who replaces a concierge who is present at the site. The operator remotely watches who arrives, grants or denies access to cars that don’t have a pass registered for them,” explains Nikolai Kulikov. “For small communities, it’s a common thing to not have a concierge of their own but employ the services of some third company.”

According to the project’s founders, the startup will continue improving their product based on the experience they’ve gained in the course of three years.