ISJ Exclusive: Intelligent security at the edge

International Security Journal catches up with Alex Osypov, CEO & Founder, FF Group.

With years of experience providing intelligent security solutions at the edge, FF Group has been working tirelessly to support traffic, parking management and access control projects by designing powerful, sophisticated and cost-effective technologies.

Developing its algorithm of image recognition for ARM, VPU and GPU-based platforms from Ambarella, ARTPEC, Wisenet, Qualcomm, NVIDIA and Intel, FF Group company is offering the security sector intelligent and efficient out-of-the-box solutions. Moreover, the company’s data processing platform, FF Data Factory, is creating convenience and simplifying large data flows for the transport sector in real time.

To find out more about the history of FF Group and how the company is innovating this market, International Security Journal spoke exclusively with CEO & Founder, Alex Osypov.

Building a bridge between mathematics and business

16 years ago, a lot of FF Group’s core team were part of an outsourcing company that worked in hardware development for the American market. As a result, the company got much of its experience in electronic design quite early on and, after several years, decided to then make the switch from selling, not just people hours, but products.

“Because of our experience, we could develop pretty much any piece of hardware we liked – we decided to jump into the CCTV market, producing amplifiers which were needed at the time in the analog video market,” explained Osypov. “We thought, ‘okay, this is kind of easy to do’, so we started to produce it.

“By 2011/12, we were selling hardware without any piece of software, however. So, we stopped, reconsidered and then added smart technology; we started to programme controllers and at this time video management system started to become mainstream. We were introduced to how to decode video, how to get frames and we developed really good retail-orientated video management systems, including hardware. That’s how we were established in the CCTV, video surveillance market; implementing our mathematical and engineering technology.

“We added licence plate recognition, something that was needed for several products. We started to develop recognition and we hired more staff with PhDs, those from a scientific background. What we discovered was that this was nothing like just building products. What we did – and I would say the real FF Group started at this time – was build a bridge between mathematics and business.

“For those with a scientific background, it’s not easy to earn from that, nor is it for those in commercial who are a bit far removed from science – but, because of the balance of our team, I would say that we found a way to monetise mathematics.

“During this period, everybody was switching to IP video surveillance and our idea was to therefore start with the Asian market which was accelerating at a rapid pace – why wouldn’t we bring all that recognition to cameras? It sounded crazy and it required a lot of mathematics, but eventually we managed it. We couldn’t use any normal computer vision technology like edge detection, so we developed almost everything from scratch (including basic math operations) to be fast on-camera.”

This became a real niche for FF Group and, at the time, the company was probably only one of several in the whole world, including big vendors, who were making such significant strides in this space. “Six-to-seven years ago, we fully switched to edge and we stopped any other business,” continued Osypov. “The focus of the FF Group was only to produce at the edge. Today, 95% of our company income comes from there.”

Building data refineries

FF Group’s work in the vehicle and vehicle-related sectors covers everything from private garages and parking lots to cities and interstate connections. Ultimately, if there are cars moving, you can use the company’s technology.

“I would call this a small vertical slice of pie from a very big sector,” Osypov said. “If you have a camera, you may need results of recognition, let’s say, to help you open your gate or barrier automatically or to analyse the movement of people across Europe.

“To be a prosperous company across these verticals is to bring your own products to the global market. Our core market was initially Europe. Then, we added South Africa, following with Australia, New Zealand and some South Asian countries, which is one of the fastest growing markets in our business. In late 2022, we will add seven countries from the Middle East. Now, our key focus is North America and Latin America.

“We are generating metadata at the edge and, with this, the next development of our business is looking at the place where we manage this data. The common analogy is that ‘data is the new oil’ – but today, it is not about who is dragging the oil from the earth, it’s who is building the refineries. It is all about how we can manage data and how we can generate insights out of data results to produce something useful for businesses and industries.”

Smart applications

Whilst FF Group’s key customers are major camera vendors like Axis Communications and Hanwha Techwin, on the platform side of things, it works with the likes of Azena as well as global VMS providers such as Milestone.

“A core feature of the business is to design high quality edge product with quick and accurate recognition for mass market. Companies like Axis and Hanwha are a great fit for us because they have a big presence in commercial markets and there’s always cameras available for everybody, from garages to city projects.

“For example, an end user can buy a camera from the distributor supply chain and just one camera can connect your garage or door controller. The next level up from this is parking territories or residential areas which require controlled vehicle access. The next level up from this we are involved in, however, is urban solutions. This could mean that a client needs a network of data sensors which can control traffic flow in a city. You can measure vehicle flow for municipalities and police cases.

“Finally, I would say the next level is the idea of a smart country. You may need to control transit flow, from south to north, or distinguish what is local traffic compared to transit traffic – for that, you need a network of measuring edge devices.

“Vehicle make and model recognition (MMR) is certainly newer than LPR, but the latter is well established. But, what’s our input to these markets? I would say it’s that we started at the edge. We have developed algorithms to use 100% of the possible capacity of CPU or GPU of a camera. Our technology has been designed and developed at the edge. I would say, technology wise, we are almost real time on the camera side of things.”

“When looking at cost-effectiveness, there are two stakeholders to consider from a business perspective. The first is the end user – ultimately, they want to buy modern technology, which provides more accurate, faster data, that is easy to install and use; they don’t want to have to worry about the complexities of the software, they want to buy products at an affordable price. The second is the vendor. If we can produce smart technology on a commercial platform, we can reach mass markets more effectively, regardless of the niche we are serving.”

So, where is FF Group heading? What is the next step? Currently, the companies are generating a lot of accurate metadata. In one month, for instance, they will have terabytes of video content and millions of events coming from multiple camera feeds. The FF Data Factory is a place to work with all this data.

“Imagine the work of Microsoft Excel table, but only for CCTV data – that’s FF Data Factory, a huge, really fast database and platform where you can collect, synchronise and analyse data from the heavy traffic flow. And, moreover, it configures individual business logic (software) in one day without a programmer. We are not writing code, we are just reconfiguring software. We are adding new data facilities to the market. I would say this combination of smart and scalable sensors is a really good step forward to help further develop the video surveillance market.”

To find out more information, visit: ff-group.org

This article was originally published in the special September show edition of International Security Journal. To read your FREE digital edition, click here.