Video analytics (Russian market)

2022: The Future - Behind the Cloud Model of Video Analytics

On February 7, 2022, IDC announced that, with the support of Ivideon, it had prepared a report on the development of video surveillance and analytics technologies and their demand among Russian companies. The study showed that cloud services, the market of which in 2020 grew by 29.9%, in the future will give video analytics new opportunities for development.

Business needs more information flows to make decisions reliably. Despite the fact that video surveillance generates terabytes of data daily, the business does not use video data 100%. It is energy-intensive and inefficient to manually process arrays of such data: according to an IDC study conducted in April 2021, 90% of video data is not used by companies due to lack of resources or tools to extract them.

A video analyst comes to the rescue, who shows confident growth. Analytic algorithms simplify the handling of such data and make them useful for business. Thanks to analytics, video cameras can be used not only for security, but also for face recognition, object counting and statistics. It is expected that companies will increasingly implement such solutions to improve customer service.

The growth of demand for video analytics solutions in the future will also be influenced by the growth of the global software market, which, according to IDC, will grow by an average of 2020-2025 in the period from 15.5%. The development of hybrid networks, as well as advanced analytics using big data and machine learning, will also contribute to the active introduction of video technologies.

At the same time, the future of video technologies is behind the cloud architecture, which has a number of advantages before storing data in its own data centers of companies. Cloud services are the fastest growing segment of both the Russian IT market and the global one. According to IDC, in 2020, the domestic cloud services market grew by 29.9%.

Why the future is behind the cloud-based video analytics model:

• Ability to test hypotheses to improve business processes. Companies do not need to develop technology solutions themselves - you can use existing services.
• Use of modern algorithms for video analytics tasks. The cloud service provider guarantees the provision of updated algorithms and latest versions of the product and is responsible for monitoring and constantly updating mathematical models.
• Continuous enrichment of datasets. Cloudy provider accumulates a large amount of data, which allows you to constantly improve training neural networks and analyze data more accurately. Full datasets are required to recognize objects - the fuller the dataset, the better the recognition algorithm will be implemented.
• The cloud provides excellent scalability and reliability of storage without the need to add hardware in the field and, as a result, increase the cost of the solution.
• The cloud-based video surveillance and analytics service supports the operational standards of companies operating through a partner network or a franchise model, allowing you to control the quality of services provided.

To leverage video surveillance and analytics, IDC recommends that business organizations:

• Set clear goals in advance for the use of video analytics. This will help build expectations and avoid unfulfilled plans.
• Avoid complex turnkey solutions whenever possible. Custom solutions require more investment and ongoing infrastructure support. Often, ready-made analytics solutions can be tailored to your query without creating new scenarios.
• Leverage platform solutions that enable business scalability and easy-to-configure data collection.
• Give preference to the cloud for data storage and analysis. This will save the cost of technical resources and will not overload the network.

Competent customer service and operational efficiency are the main factors for the success of the business, and sometimes the factors for its survival. Therefore, companies need tools to better understand their target audience and improve their business processes. One of these tools was video analytics. Cameras equipped with video analytics modules not only provide security at the site, but also consider visitors, analyze their mood and help manage business metrics that affect sales. And this is only a number of tasks. At the same time, companies no longer need to invest in a large amount of equipment, software and licenses, installation and maintenance of all this. To work such a service, you need a camera and Internet access. This gives managers the opportunity to manage all processes remotely through a single window, "said CEO Ivideon Andrei Yudnikov.

2020: Market growth by 14.2% to 11.1 billion rubles

The Information and Analytical Agency TelecomDaily July 8, 2021 presented the results of a study on the state of the Russian video analytics (VA) market.

According to the study, the volume of the VA market in the Russian Federation at the end of 2020 amounted to 11.1 billion rubles, which is 1.4 billion (14.2%) more than in 2019. The fastest growing segment is called cloud VA with a market volume of 1.9 billion according to the results of 2020.

According to the results of 2020, the first place in the cloud VA market was the cloud platform operator - Ivideon with a market share of 21.9%. It is followed by a platform from the telecom operator Rostelecom (15.5%). Closes the top three "Sinesis" (6.7%).

The main areas of application of video analytics technologies based AI on in 2021 are (retail 15%), production, transport and (13% each logistics) and (10 ENERGY INDUSTRY %). According to the forecast, by 2028, the dynamics of the cloud VA market will reach 25.9 billion rubles, ahead of the market as a video surveillances whole by 10-12%.

More than 97% of the revenue of the cloud VA market falls on the b2b segment. b2g accounts for 1.2%, b2b2c - 1.3%. The low share of the public sector is explained by the fact that data from video cameras are written mainly to the data processing centers of government customers, and not to the cloud of the service operator.

The entry into force of the law FZ-152 led to the need to certify its own system, which requires a lot of time and costs to ensure the necessary security, which is why companies are increasingly using operators "clouds" certified according to FZ-152.

2019-2020

Overall Market Status

Scientific research in the field of video analytics, computer vision and artificial intelligence has been conducted in Russia since the 2000s on the basis of research centers and universities. To date, video analytics algorithms have been used mainly for the following purposes:

• event detection;
• counting of visitors;
• recognition of dangerous objects;
• face recognition and identity identification.
• ensuring security at various facilities (protected areas, transport, as well as at state facilities).

According to Marina Ivanchenko, Deputy General Director for Strategic Development of Center 2M, video analytics is still far from mass implementation in Russia. But in the industry there are three important trends that will strengthen each other and, in the end, will provide the market with rapid growth and synergy of certain areas.

• Firstly, this is an increase in confidence in smart solutions. The emerging video analytics market is focused on successful pilot projects. The more such cases, the more interest in the industry will be shown by new players.
• Secondly, this is an increase in the number and quality of cameras and sensors. Large cities are entangled in a network of cameras, but most devices are outdated and not suitable for implementing video analytics solutions. As the technical fund is updated, the number of available data use cases will also increase.
• And the third trend is the so-called Sharing economy, one of the drivers of video analytics in smart cities. To successfully implement scenarios, camera owners must be prepared to allow other infrastructures access to the video stream.

Center 2M believes that sharing information is necessary first of all:

• emergency services - to increase the level of security;
• tenants of retail space - to analyze the cross-country position of retail outlets;
• retailers - to develop personalized marketing based on video biometrics and data from CRM;
• city administration - to monitor and generate statistics on the workload of service centers.

Of course, when sharing data, it is necessary to take into account security requirements and legislative norms that limit the use scenarios of both personal and depersonal data.

In general, Marina Ivanchenko from Center 2M believes that it is still difficult to say which barriers - legislative or economic - can impede the development of the market. The regulatory framework in the field of video analytics remains unresolved.

According to VK (formerly Mail.ru Group), for the video analytics market to mature, 10-15 industry cases should appear. Up to this point, the business will be helped by support in the form of consulting, up to the formulation of needs for the client^[1]. Such projects are becoming a reality.

Mayor Sergei Sobyanin announced on his VKontakte page that "the facial recognition system for the search for criminals has already been introduced in the metro, and this year (2019) will appear on street cameras. And it will be cool: the criminals will stay away from Moscow, here they will no longer hide. " According to him, currently the city's video surveillance system has 167 thousand cameras - they are installed in yards, entrances, parks, schools, polyclinics, at trade and construction facilities, as well as in the halls of executive authorities and other public^[2].

According to J'son & Partners Consulting, in Russia there is an active penetration of facial recognition technology (Facial Recognition). It increased its share in the total volume of the Russian biometric market to almost 50%, showing significant 106.7% growth (CAGR) in the period 2015-2018^[3].

X5 Group uses the "gepcheck" procedure: scanning all goods in the department and comparing with the balances in the warehouse and planograms^[4].

In addition, video recognition can help you understand how correctly the store area is divided into separate zones. Using video analysis, you can track how many people entered a particular section during a day or week. This information can be used to optimize the location of racks in the sales room.

At the international industrial exhibition INNOPROM-2019, Krok presented a video analytics solution for labor protection and industrial safety (OtiPB). The system is able to recognize faces, monitor the situation on the territory of the facility, monitor compliance with safety and labor safety standards^[5].

According to Ivideon, about 15 million CCTV cameras are used in Russia, with 22% of them using video analytics: facial recognition, counting unique visitors, queuing analytics and layouts for retail ^[6]

Recently, the direction of predictive video analysis has been developing. Such technologies are being tested in retail, to trade including in the Lenta retail chains and. X5 Retail Group More and more requests for such functions come from production enterprises that improve the quality of production, conduct automated control over the safety of equipment and the work of^[7]

The industrial video analytics system of the Center 2M company CenterVision contains a neural network and can track safety violations. The system transmits data on violations, monitors the state of the territories, and also warns of a cluster of people, working in real time. Since 2018, the company has launched a number of pilot projects at large enterprises and in municipal institutions of the Russian regions.^[8]

At the Russian-British Business Forum in London in 2018, the Center 2M company presented a development in the field of machine vision CenterVision based on the IBM PowerAI Vision platform, which allows you to quickly implement the system in production and ensure effective object recognition. The Center 2M industrial video analytics system uses neural networks to recognize objects at production facilities with an accuracy of 98%. At the same time, it takes only 1 day to train the neural network from scratch, to train a new scenario - 30 minutes^[9].

Spetslab has developed the I Rain system, which allows you to automatically distribute the image of a suspicious person to all security terminals at the^[10] enterprise^[11]

California research group Market Intelligence Data in 2019 included Synesis from Skolkovo in the list of leaders in the video analytics systems market, along with companies such as Axis Communications, Cisco, Bosch Security Systems^[12].

Moscow coffee shops OneBucksCoffee and Pravda Coffee are introducing a cloud-based customer facial recognition service based on Ivideon's solution. In the Pravda Coffee network, the project is piloted in nine institutions with the prospect of introduction in all 60 operating points. In OneBucksCoffee, the service operates in four coffee shops, but the company also does not exclude the possibility of deploying a video identification system for visitors to the entire network^[13].

Developer of computer vision systems VisionLabs (holder of a grant from the Innovation Assistance Fund, resident of Skolkovo; 25.07% - at Sberbank, 25% - at the Sistema_VC AFK Sistema venture fund) became a partner of Yandex.Taxi to launch a driver fatigue and attention monitoring system. Yandex.Taxi in the next three years plans to invest in taxi safety technologies in the amount of 4 billion rubles.

In addition to VisionLabs, in Russia, facial recognition technologies are being developed by Vokord, the Center for Speech Technologies (CST) and NtechLab^[14]

The forecast of the video surveillance market in the Russian Federation until 2022, part of which is the video analytics market, according to MGTS, is shown in the figure below.

Ivideon estimates the video surveillance and video analytics market in Russia in 2018 at 40.8 billion rubles. The volume of deliveries of video surveillance cameras in 2018 amounted to 10.8 billion rubles, the volume of the niche of cloud video surveillance services - 4.1 billion rubles^[15].

Video analytics market estimates by its participants

Market Volume Estimate

According to a survey of companies, we can conclude that the video analytics market in Russia will grow quite quickly. According to TAdviser, the average annual growth rate of CAGR will be 20.4%. In 2019, the video analytics market in Russia is estimated at 18.79 billion rubles. For 2025, the projected market volume is 51.75 billion rubles.

Many companies, however, indicated that market growth indicators will depend on the overall state of the economy of the Russian Federation and the world, since the introduction of video analytics technologies does not pay off explicitly. That is, real numbers can be very different from the predicted ones.

According to Maxim Mirolyubov, director of the video analytics department of Center 2M, such a rapid growth is due to the fact that with the beginning of the application of deep learning methods to the video stream received from cameras, passive video surveillance has turned not only into a method of collecting and analyzing data, but also into a method of automated process control with the ability to immediately respond to events and situations of interest.

A few years ago, such technologies would be very expensive and time-consuming, but now we have solutions that can implement almost any necessary video analytics scenario using artificial intelligence, "said Maxim Mirolyubov, director of the video analytics department of Center 2M.

According to him, as of February 2020, the company has about 30 scenarios for building intelligent video analytics systems. Mirolyubov emphasized that they account for 80% of all video analytics requests available on the market. The most popular scenarios involve perimeter surveillance, biometric identification, as well as queue control and the use of personal protective equipment.

Other estimates

In the course of a study conducted by TAdviser in 2019, a survey of its participants was conducted, which included developers and integrators of video analytics solutions.

The results of the survey are given below.

The most popular types of video analytics on the Russian market

The following analytics were also mentioned as popular types:

• Analytics for retail (hot and cold zones, product layout);
• Analytics for labor protection (detection of dressed helmets, workwear, behavior, etc.);

Incident Detection, according to the survey participants, although it is in demand, however, there are no good solutions for practical implementation due to technological problems, as well as due to the inability to respond to the incident in practice in time.

Demand for video analytics by sectors of the Russian economy

The most promising scenario for deploying video analytics systems

According to the results of the survey, out of three possible deployment scenarios for video analytics systems

• Cloud (SaaS,...);
• In the corporate system (On-premise);
• Mixed (Hybrid);

90% of the companies surveyed indicated a mixed (Hybrid) deployment scenario, when the main system is deployed on local enterprise resources, and some analytics functions, or video storage space, can be leased in the cloud. It is worth noting that 2-3 years ago, most enterprises indicated as the most promising deployment scenario in the corporate network (On-premise), showing distrust of cloud solutions.

In many cases, the driver of cloud use is the entry into force of law FZ-152, since certification of your own system under the law on personal data is often time-consuming and costly. The best solution in such cases is to rent resources in a cloud certified according to FZ-152.

Perspectives of innovative technologies in the video analytics market in Russia

The results of the survey showed that innovative technologies such as Big Data analysis, artificial intelligence, neural networks, machine/deep learning will be the main factor in the growth of the video analytics market in Russia.

For example, Marina Ivanchenko from Center 2M told TAdviser analysts that according to the annual Gartner rating, 4 out of 10 top technologies in 2019 occupy trends in video. These are trends such as:

• augmented analytics
• immersive experience
• smart spaces
• digital ethics and personal data security (digital ethics and privacy).

These trends owe their existence to the new reality of the digital world. There is too much data around us, and humanity needs to learn how to work with this array of information.

Automated video analytics technology helps solve the problem of data processing. 5-10 years ago, it was impossible to implement such solutions on an industrial scale. Each project required several hundred man-years and a whole arsenal of supercomputers.

Now computing power has risen to a qualitatively new level. The Center 2M company has solutions that can process all types of data using artificial intelligence based on convoluted neural networks.

Video analytics of a new level do not just meet the needs of the new environment. Innovative solutions are becoming a mandatory tool in the business sector. Competition in the corporate environment is growing, and the possession of even the most modern means of production no longer guarantees stability. We need new systems of operational situational analysis that will help better manage the business: for example, monitor the basic infrastructure or work to improve the efficiency of production security systems and make diagnostic decisions.

The company Macroscope notes that synergy of decisions will be of great importance. According to Macroscope, more and more complex solutions will appear, in which one VMS video surveillance management system will act as an operating system for situational use of possible solutions.

According to X-Com, new technologies make it possible to continuously improve analytical algorithms and open up qualitatively new opportunities for applying these solutions, increasing the return on investment in them. For example, in retail, they allow you to determine the demographic composition of visitors to retail outlets, identify the frequency of visits and the composition of the food basket of various groups of customers, optimize the layout of goods, plan and organize targeted market activities, and build effective omnichannel communication with the audience. An analysis of the mood of visitors at the entrance to the store, in the trading room and at the exit allows you to identify and eliminate the shortcomings of business processes (including those that arise only at a certain time), thus maintaining customer loyalty. Security Services can help identify potentially dangerous customer behavior and prevent misconduct.

Tevian believes that now almost all video analytics are currently based on deep-learning neural networks. Without these solutions, only the simplest and fastest methods work (part of perimeter protection methods, etc.).

NVIDIA predicts that the most promising areas of development of video analytics systems in the near future will be as follows:

• improving the quality and speed of facial recognition, which makes it possible to pass through the control point without stopping, without the need to look strictly into the camera;
• liveness check (the impossibility of using a photo or mask instead of a living face);
• recognition in conditions of poor visibility (shooting in the dark, in rain/snow/fog), and situations where the face is partially covered with glasses, scarf, headgear, etc.),
• emotion recognition;
• recognition by silhouette, clothing and gait, including from the back;
• weapon recognition, the possibility of using geotags and any additional information to exclude the possibility of forgery;
• recognition of behavioral scenarios (a person needs help, health problems, violence, etc.) in conjunction with automatic notification of relevant services.

Impact of Federal Law FZ152 on personal data on the video analytics market

According to most of the companies surveyed, from the definition established by Federal Law 152 "On Personal Data," materials with images of citizens obtained as a result of video shooting in public places and in the protected area are not biometric personal data.

Roskomnadzor recommends equipping places under video surveillance with warning inscriptions or graphic images. If this condition is met, then a separate consent of citizens to the shooting is not required. Examples of using such a video analytics mode include monitoring the quality of customer service, ensuring the safety of property, monitoring the workplace, etc. Moreover, hidden video surveillance is prohibited by law.

Interviewed companies note that according to the law, video is not a carrier of personal data, from the point of view of the law, there is an image of a certain person on the record, and no more. Such records can be stored without limitation. But if a facial recognition system is used that assigns certain identifiers to images, then the law FZ152 comes into force, and its user becomes the operator of personal data. Therefore, the storage infrastructure of such an organization must comply with the requirements of the FZ-152. But certification of your own storage system is a very protracted and costly process, so in many cases it is cheaper and easier to rent the necessary resources in a cloud that is already certified for storing personal data.

Some companies, however, believe that FZ152 has a significant regulatory impact, defining and somewhat limiting the use of video analytics systems. To actively develop the market, it is necessary to develop a separate package of requirements for video analytics systems and clearly determine what falls under the said law and what is advisory in nature.

A number of companies also believe that the Law on Personal Data slows down some of their projects to introduce facial recognition due to the fact that some customers doubt the possibility of using facial recognition without violating the FZ-152. However, this law does not greatly affect most projects, but in general it slows down the market a little.

Main Russian vendors of video analytics systems

The company "Center 2M" names the following most prominent players in the video analytics market in Russia:

• Infrastructure manufacturers and software developers (vendors)
• * Qlik (QlikTech)
  • Forecast
  • Loginom Company (formerly BaseGroup Labs Analytics)

• * SAP SE
• * IBM
• * Oracle
  • Business Automatization SPC

• * SAS Institute Inc
• * Contour Components
• * MicroStrategy
• Equipment manufacturers
• * Axis Communications AB
• * Bosch Security Systems Incorporated
• * Honeywell Security Group
• * Samsung Group
• * Panasonic Corporation
• * FLIR Systems Inc.
• * Schneider Electric SE
• * Zhejiang Dahua Technology Company Limited
• * Hikvision Digital Technology Company Limited
• * Sony Corporation

X-Com specifies the following companies:

• • Trassir (Firmware Developer)
  • Hikvision (Infrastructure Developer)
  • AxxonSoft (software developer)
  • AXIS (Infrastructure Developer)
  • Vocord (Software Developer)
  • Dahua (Infrastructure Developer)
  • Smartec (Infrastructure Developer)
  • Macroscop (Software Developer)
  • Honeywell (Infrastructure Developer)

The list of the main Russian video analytics vendors registered with the Skolkovo Foundation is presented in the table below ^[16].

Below is a summary of the main Russian developers of solutions that are related to the field of video analytics. Information is given only about companies that themselves develop and provide video analytics solutions.

Information about system integrators, suppliers, service companies of solutions of other developers (both Russian and foreign) is not given (otherwise it would be a very long list).

Center 2M LLC

The industrial video analytics system of the CenterVision company Center 2M uses neural networks to recognize objects. Using this technology allows you to recognize objects with an accuracy of 98%. Monitors the production process in real time, sends an automatic alert to the situation center, prepares reports on all facts of violation of security rules.

Advantages of the Industrial^[17]

• Reduce manufacturing errors. The accuracy of neural network recognition up to 98% reduces the probability of error to an absolute minimum;
• Instant violation notification. A response time of 2.4 seconds allows the system to instantly learn about safety violations;
• Low impact of weather conditions. The detection of objects from a long distance, regardless of rain, snow or fog, reduces dependence on weather conditions;
• Coverage of large territories with small resources. Recognition of small objects from a long distance (50-70 m) makes it possible to control a smaller number of video cameras.

The 2M Center introduces modern video analytics systems with individual settings for the customer: at production, state facilities, in educational institutions, hotels, banks, stores.

Depending on the specifics of the facility, Centr2M offers several complex solutions for industrial safety. Patented technologies will help increase the percentage of recognized objects to 98%. A successfully implemented scheme of human and machine vision joint work is already being used in regime enterprises and complex industrial facilities.

Vokord SoftLab LLC

Video analytics solutions VOCORD are being promoted to the international market (, India Middle Eastern countries states and the Persian Gulf). Currently, according to the Skolkovo Foundation, VOCORD's share in the Russian market is 10%. VOCORD^[18] Solutions]:

• VOCORD Traffic: Number recognition and recording of traffic violations. Intelligent traffic control system, operates in automatic mode 24/7 in any weather conditions. Multifunctional intelligent traffic control system with high reliability of number recognition and fixation of more than 15 types of traffic violations.
• VOCORD Tahion: Video surveillance and video analytics. Software and hardware complex for creating video surveillance systems and video analytics of any scale and format. VOCORD Tahion intelligent video surveillance is successfully used in 70 projects of the Safe City class. At emergency road sections and complex junctions, where it helps to collect important information for managing traffic flows. On highways and federal highways, where it gives alarms to the dispatcher in case of prohibited incidents and monitors the situation at payment points. In factories, warehouses, hypermarkets and offices, where it recognizes objects and actions, helping businesses solve various problems. The VOCORD architecture allows you to choose functionality, flexibly configuring the system for the desired purposes.
• VOCORD ParkingControl: Smart parking. System for automation of the pass mode based on recognition of car numbers and control of vehicle entry/exit. Allows you to speed up the work of the guard at the checkpoint or fully automate the access mode, minimizing the risks of the human factor. The system logs all actions in the control area, allows you to add new permissions in real time, keeps records of unauthorized actions, and also uploads statistics on transport on site to flexible reports. In case of disputes, the system archives the video recording according to the specified parameters.

Video Intelligence LLC

Video Intelligence is a predictive video analytics system for various branches of the market, recognizing objects and people, based on algorithm building technology, using nuclear physics approaches to effectively work with weak signals and machine and deep learning algorithms^[19].

DiSiCon LLC

The "Forest Watch IS PAK" system of DiSiKon LLC is a software and hardware complex for operational monitoring and detection of forest fires, which allows you to remotely record forest fires and determine the coordinates of the fire center with high accuracy.

The basis of the system is the "Forest Watch" software of its own development.

Synes

The Kipod platform of Synes LLC is a cloud-based video analytics platform for fuel and energy complex, construction, logistics, production and other areas, which allows automated monitoring of remote facilities. The platform uses technologies for ranking video data and a wide range of video analytics modules with confirmed accuracy indicators.

Kipod effectively uses a bandwidth-limited communication channel due to the Synes video data ranking technology. It guarantees the effective transfer of data not by reducing their quality, but by determining the priority of a particular video fragment.

LLC Videomatrix

The VideoMatrix project is a software platform for tracking the products of industrial enterprises with conveyor production, solving the problem of continuous accounting and quality control of products through the use of unique algorithms for processing a single image based on an array of video cameras.

Based on the theory of pattern recognition, VideoMatrix algorithms solve the problem of tracking the movement and storage of products in conditions of limited and variable visibility. Products can be stored in unstructured form, follow with an arbitrary interval, orientation, in several flows. It has a low implementation cost and flexible functionality due to the use of video cameras instead of highly specialized devices. Provides end-to-end tracking of products without the use of special markings.

ElVIS-NeoTech

Orwell 2k video surveillance system is designed for automatic detection and classification of targets and situations, real-time transmission of video information to the operator. Computer vision algorithms used in the system are the development of ELVIS-NeoTech. Orwell 2k computer vision algorithms minimize false positives and detect situations and targets in a complex interference environment.

Macroscop

Macroscop is a program for managing video surveillance systems based on an open platform, a wide range of video analytics modules and network video recorders (NVR)^[20].

Integra-C Consortium

The Integra-Video video surveillance system is designed for continuous visual monitoring, video recording and motion detection in protected areas at objects of any scale^[21]. The system allows you to build a distributed structure of any complexity with the connection of an unlimited number of servers, video cameras and users. Video surveillance configuration is carried out through one (central) server, which transfers all settings to other (slave) servers. You can also configure each server individually.

List of analytical functions of the system:

• Motion detector
• Object motion detector
• Crossing a Line in a Forbidden Direction
• Enter/Exit Zone
• Loitering
• Fight Detector
• Running Detector
• Tracker
• Autotracker
• Image Quality Detector
• Retained/Removed Items Detector
• Smoke detector
• Crowd Detector
• Human Count Detector
• Sound detector
• Activity Zone Detector
• Queue Detector
• Open flame detector
• Traffic congestion detector
• Monitoring of traffic light equipment
• Search by Color

Computer vision systems

The Computer Vision Systems company develops solutions in the field of machine vision, image and video analysis, based on research from the Department of System Programming of St. Petersburg State University together with the IT company LANIT-TERCOM^[22].

TRIDIVI (3DiVi Inc.)

TRIDIVI (3DiVi Inc.) develops software for computer vision, Artificial Intelligence and deep learning, focuses on the areas of Security and Analytics, Augmented Reality (AR), the Internet of Things (IoT) and robotics^[23] products:

• Nuitrack - body tracking and gesture recognition software
• TVico - Interactive Android computer
• Face SDK - Facial Recognition Software
• Seemetrix - video analytics for advertising displays

Intelligence Retail

Intelligence Retail offers digital merchandising (retail) solutions for mobile phones and tablets, stationary cameras, robots and IoT^[24] devices].

ITV|AxxonSoft

ITV AxxonSoft offers development of video analytics, which are used in many areas.^[25]

• Video analytics of the Intellect software complex: a video detector system and an intelligent search system in a video archive.
• Video analytics of the video surveillance system Axxon Next: 10 video detectors, interactive detector configuration, intelligent search system in the video archive, automatic reactions to events, etc.
• Car number recognition module ("Intelligence"): recognition of state registration plates of vehicles.
• Facial recognition module ("Intelligence"): identification of a person by video image by comparison with reference images in the database.
• Module for monitoring the characteristics of traffic flows ("Intelligence"): determining the types of vehicles, calculating the intensity of traffic in a given area, collecting statistics and more.

Video Analysis Technology (Tevian)

"Video Analysis Technologies" (Tevian) was founded in 2010 by graduates of the Computer Graphics and Multimedia Laboratory, Faculty of Computational Mathematics and Cybernetics, Lomonosov Moscow State University, specializing in applied research and development in the field of computer vision and video analytics, security systems, audience analysis, intelligent search in video archives, interactive advertising, augmented reality systems, etc. In 2013, Video Analysis Technologies received the status of Skolkovo resident with the project 'Automatic annotation of video surveillance data for business intelligence and security systems^[26]. Key Technologies:

• classification of a person's sex and age;
• human identification;
• anthropometric points of the face.

NtechLab

NtechLab develops solutions in the field of artificial neural networks and machine learning. Facial recognition algorithms were recognized as the best in the world by independent third-party tests^[27]

Intellojik LLC (TeleMD)

TeleMD is a software platform for diagnosing and analyzing the risks of developing diseases based on mathematical models of patient representation using artificial intelligence technologies. The TeleMD platform uses a combination of three main technologies:

• Artificial intelligence technologies for processing medical images and combining heterogeneous medical data into a vector encoding all medical patient data;
• Medical ontologies for the preparation of mathematical models and the interpretation of the results of the use of artificial intelligence technologies;
• A results presentation system that simplifies the evaluation of the validity of the mathematical model by a specialist physician.

TeleMD provides APIs for integration with clinical, telemedicine and other information systems.

PAVLIN TECHNO LLC

PAVLIN Techno LLC develops solutions in the applied fields of video analytics^[28]:

• Automatic construction of digital 3D models of terrestrial objects and the earth's surface based on laser scanning data, including the construction of height maps, the identification of roofs of buildings, the selection of extended objects (power lines) surrounded by vegetation;
• Photorealistic 3D reconstruction of buildings and reliefs using panoramic optical stereo photography;
• Detection and recognition of faces, logos. Detection of foreign objects on X-rays in flaw detection tasks. To implement these projects, the original technology of detecting and recognizing arbitrary objects and scenes in the video stream and in photographs was invariant to scale, rotation, position and moderate brightness and color distortions;
• Development of autonomous domestic robots with intelligent on-board control system;
• Development of computer vision and data analysis algorithms for unmanned aerial vehicles (UAVs);
• Development of an object recognition system on radar imagery;
• Development of UAV recognition and tracking system on video.

Biometric Systems

Loop-Stream is a complex of biometric recognition of a person from a face image in a video stream in real time^[29].

• Reception of RTSP and HTTP video streams from CCTV cameras and other sources;
• Detection in the video stream of persons and their search in databases;
• Saving recognition events in system archives;
• Complex equipment control;
• Automatic computational load balancing;
• Create failover clusters;
• System integrity and recognition quality control;
• Connection of response and positioning system;
• Open API for integration with other systems.

Stilsoft

Stilsoft Trading developed the Security Corridor and CerbeRus complexes, which provide optimal lighting conditions and minimize background interference at the time of recognition. These complexes work together with the AIS "Synerget Search" automated information system, which allows you to combine data used for facial recognition with personal, anthropometric and other operational data, which ultimately allows you to narrow the volume of the recognition base and get much more relevant data that best fit the operational orientation^[30].

Bell Integrator

Bell Integrator's Bell.One intelligent video analytics service allows you to analyze a large stream of visual data, enabling you to effectively respond to potential threats. The video analytics service includes digital processing and automatic analysis of video streams - in real time or recording - in order to obtain semantic information about the content^[31] video sequence].

"Light Technologies" (Deus Company)

Deus developed the video analytics module^[32] and a special interface in the ME6 platform, which allows:

• Create the required number of detection zones (of any shape) in the camera scope;
• Snap to created lighting group zones;
• Determine the number of moving objects in the frame;
• Define the motion path of moving objects;
• Script lighting groups based on the path of objects.

VideoNet

The VideoNet system comprises intelligent detectors which automatically detect and respond to suspicious and dangerous events. For each camera, different combinations of detectors and detection zones with individual parameters are created. The system of reactions to the occurrence of events from various subsystems, sensors, equipment allows you to create complex scenarios to warn the operator. VideoNet informs about the event, attracts the attention of the operator and minimizes the time of his reaction to the event^[33].

Vizillekt

Visillect provides a wide range of services in the field of research and creation of visual systems: consulting services and pre-design surveys, research and development, the creation of software products and software and hardware complexes, as well as research on the properties of existing^[34] systems] in the following areas, which use end-to-end video analytics technologies:

• Technical vision algorithms;
• Streaming recognition systems;
• Video analytics;
• Stereo and 3D reconstruction;
• Color vision and multispectral image analysis;
• Visual navigation and odometry, visual memory;
• Adaptation and self-calibration;
• Augmented reality technologies;
• x-ray Imaging methods (including tomography);
• Use visual properties in man-machine interfaces.

PJSC NIPS Novosibirsk Institute of Software Systems

The video surveillance system "Video Guard," developed by specialists of PJSC "NIPS" works with both digital and analog video equipment and allows you to conduct video surveillance over the Internet from anywhere in the world. The Videogard system is implemented on IP-video surveillance technologies and allows you to solve a range of tasks for continuous video monitoring and monitoring the health and integrity of the video surveillance system itself^[35].

LARGA Group of Companies

LARGA Group is developing and implementing video surveillance systems with advanced analytical capabilities based on its own platform LARGA.Videoserver. To store and process the video stream, a private cloud solution is used, which is a software and hardware complex: a server with installed software and a storage system that are located at the customer^[36].

Kraftway

Kraftway Smart Video Detector - professional cross-platform software video surveillances with video analytics^[37] ON supports operating system, Windows(Linux Debian, Ubuntu) and. Astra Linux SE KSVD software is designed for video surveillance of objects of various scales:

• Small: from 1 to 50 cameras - house, office, store.
• Average: from 50 to 300 cameras - office building, factory, hypermarket, enterprise.
• Large: more than 300 cameras - city district, city, several cities.

Vipax LLC

Vipax LLC offers the video analytics program "Domination Video Client" 2.2^[38]

• a report function is available for the Heat Maps module with the ability to export to PNG format;
• support of loud sound detector events;
• a filter on the received events in the Video Analytics Events widget and in the search for video analytics events;
• Save the event from the Analysis Events widget to PNG or GIF format.

Vizorlabs

VizorLabs is one of the leading Russian suppliers of video analytics systems for industrial applications, a resident of Skolkovo Technopark. The company develops technologies for recognizing and automatically processing photos and video images using neural networks.

Platform video analytics solution VizorLabs is a software and hardware complex with artificial intelligence for controlling technological processes; accounting and quality control of products; staff monitoring; industrial safety monitoring; flaw detection and telemetry.

The complex consists of a computer vision system, a media server for processing and storing images, as well as user interface modules, a reporting and alert generation module, and a neural network training module.

The computer vision subsystem is a cascade of neural networks (detectors) for solving specific problems: motion detection, detection of people and objects, counting objects, determining the position of objects relative to each other, etc.

In total, the library VizorLabs more than 75 ready-to-use detectors specially designed for typical industrial processes. The VizorLabs solution also features a unique function of recognizing actions and the ability to independently train neural networks by the customer.

The VizorLabs technology ensures the accuracy of detecting objects from 95% in all angles, with the angle of inclination of the camera relative to the floor up to 70 degrees, at a distance of up to 25 m, on a video stream with a resolution of 640 * 480 pixels. Video flow from 1000 + cameras can be processed simultaneously. The solution is implemented server in and, cloudy or SaaS-version. Based on the platform solution of video analytics, such industry products as the automatic safety monitoring system VizorLabs Health & Safety and the automated system for monitoring compliance with the sanitary (mask) regime in public places VizorLabs Anti-COVID are implemented.

• 1 XLIFFService: Error in XliffFile2XliffString method.

Company VizorLabs is a Russian supplier of video analytics systems for industrial applications, a resident of the Skolkovo technology park.

VizorLabs develops technologies for recognition and automatic processing of photos and video images (computer vision) using artificial intelligence and neural networks to solve applications in the field of automation of production processes, flaw detection and telemetry, in access control systems and industrial safety.

The platform solution of video analytics VizorLabs based on artificial intelligence is a cluster of jointly working neural networks developed to solve specific problems: motion detection, detection of people and objects, determining the position of objects relative to each other, counting objects, etc.

Combining detectors in different configurations allows you to adapt the system to solve a wide range of computer vision tasks: process monitoring and quality control, industrial safety control and personnel management. In total, video analysts VizorLabs more than 75 ready-for-use detectors in the platform solution library (data for December 2020).

The VizorLabs technology ensures the accuracy of detecting objects from 95% in all angles, with the camera inclination angle relative to the floor up to 70 degrees, at a distance of up to 25 m, on a video stream with a resolution of 640 * 480 pixels. Video flow from 1000 + cameras can be processed simultaneously.

The solution VizorLabs distinguishes between the functions of action recognition and intuitive additional training of neural networks by the customer, wide capabilities of hardware and software integration with other systems. The solution has software- and hardware components and is implemented in both server and cloud versions, or SaaS version.-- >

2017: There is too much marketing in the video analytics market

Scientific research in the field of computer vision and artificial intelligence has been conducted in Russia since the 2000s on the basis of research centers and several large universities.

In Russia, until recently, video analytics algorithms have been used mainly for detecting events, counting visitors, recognizing dangerous objects and identifying persons in order to ensure security at various facilities: protected areas, transport (airports, railway transport, recognition of car numbers for the traffic police), as well as at state facilities.

Modern developments in the field of video analytics are capable of solving a large range of commercial problems. Algorithms can collect and analyze important marketing information in real time (counting people and transport, analyzing queues, monitoring the activity of people in individual zones). The high accuracy and reliability of data obtained as a result of the operation of video analytics systems is confirmed by the extensive use of algorithms in business.

Vitaly Kuznetsov, Managing Partner of Office Anatomy: In the market of machine vision or video analytics there is too much marketing from developers, little experience with integrators and high expectations from the new product from the customer. As a result, the developer boasts a newly minted, but still far from ideal product, the integrator sells it to the client, and the customer, heated by pictures from Hollywood blockbusters with spectacular scenes of face recognition and detection of criminals, receives one disappointment in the remainder. Everyone is too interested in getting a revolutionary product that will irrevocably change the security market.

See also

• Video analytics systems - Catalogue of video analytics systems and projects

• Video surveillance systems - catalogue of systems and projects
• IB - Biometric identification - catalogue of systems and projects
• VisionLabs (VisnLabs) - VisionLabs Luna
• Synesis Cephei Video Analytics Complex - Synes
• Synes: Video Analytics Device - Synes
• Kipod Cluster - Synes
• CasRetail - Synes
• VisorJet with video analytics for a rotary camera - Elvis-NeoTech
• ELISE Semantic Processor (ELVEES Image Semantic Engine) - Elvis-NeoTech
• Video Intelligence: Video Analytics System - Video Intelligence
• Video Synopsis - BriefCam
• R-Style: PRIME Video analytics - R-Style Computers
• Business analytics: Video analytics for shopping centers - Business analytics
• How computer vision systems change logistics and medicine
• Vocord (Vocord)
• Yandex Data Factory
• Cognitive Technologies
• Macroscop (Satellite Innovation)
• Aggregator
• Integra-C
• 3iTech (formerly 3i Technologies)
• LogistiX (Logistics-Tech)
• Intelligence Retail
• Wireless Analytics (WiFi Analytics)
• Wireless analytics in Russian retail
• Visitor counting technologies
• Retailbase: WiFi analytics - NPO Analitika
• VizorLabs (Vizorlabs)

Notes