Video Analytics (Global Market)

2023: Global Intelligent Video Analytics Market Size Up 20% to $4.95 Billion for the Year

In 2023, the global cost of solutions for video analytics based on artificial intelligence reached $4.95 billion. This is about 20% more compared to 2022, when expenses amounted to $4.11 billion. Key growth drivers include technological advances, increasing public safety requirements, real-time monitoring, and the adoption of edge computing. Market trends are addressed in the Market Research Future survey published in early September 2024.

The growth of demand for AI video analytics is largely facilitated by the development of machine learning and neural networks. These technologies improve the accuracy and efficiency of video processing, as well as solve complex problems such as object and face recognition, behavioral analysis, etc. As a result, it becomes possible to develop more reliable and functional systems capable of detecting and classifying events based on video data.

Worsening security issues in public places and private areas further stimulate the need for AI video analytics tools. Organizations around the world are investing in advanced systems to improve surveillance capabilities, situational awareness, and timely incident response. Intelligent video analytics platforms use complex algorithms to analyze the video sequence in real time, detect and classify suspicious actions, objects and faces. By fully automating the monitoring process, this analysis helps businesses improve safety and minimize response time to incidents. In addition, new opportunities for deploying AI video analytics are creating smart cities and intelligent transport platforms: improving public safety, monitoring traffic patterns, detecting accidents and providing drivers with information in real time are becoming possible.

According to the deployment model, the AI video analytics industry is segmented into cloud and local. In 2023, the costs in these areas amounted to $2.68 billion and $2.27 billion, respectively. The cloud model allows companies to access services without having to invest in expensive hardware and software infrastructure. On the other hand, the local model provides more control over data and security. The authors of the study cite retail, healthcare, manufacturing, transport and public administration as key areas of application for intelligent video analytics. In terms of implementation, safety and surveillance are the leading segment with a share of about 35% in 2023.

Significant market players include, Genetec Avigilon,,, Hanwha Techwin,, Uniview Pelco Panasonic IndigoVision,,,,,,, and Dahua, Honeywell. Geographically, Bosch Security Cisco Verint Systems Hikvision Axis Communications Dallmeier he leads, followed by North America. At the same time, Europe the Asia-Pacific region is experiencing the highest growth rates due to the introduction of video analytics in retail, transport and to trade production.

At the end of 2024, revenue on the global AI video analytics market is estimated at $5.97 billion. Market Research Future experts believe that in the future, the average annual growth rate in complex percentages (CAGR) will be 20.5%. As a result, by 2032, costs globally will reach $26.54 billion.^[1]

2019-2020: Global Market Situation and Forecasts Overview

The global video analytics market is showing rapid growth thanks to ever-declining prices for high-resolution video cameras. Even small and medium-sized businesses are now quite capable of purchasing an autonomous video surveillance system with at least elementary video analytics functions. IP cameras have high resolution and remote access capabilities, both over the Internet and through a distributed corporate network, which is a prerequisite for the operation of video analytics systems.

Video Analytics Systems and Services Market Growth Prerequisites

The cost of IP equipment (cameras, switches, DSS servers, etc.) is constantly decreasing, which leads to its widespread distribution, and at the same time to the distribution of video analytics programs and applications. If ten years ago IP technology for video surveillance was considered expensive, now the total cost of owning TCO (Total Cost of Ownership) IP video surveillance systems has been reduced to values ​ ​ that contribute to the rapid distribution of software and video analytics services.

The price-performance ratio for IP cameras shows compliance with Moore's law^[2]. The price of high-resolution cameras has dropped from 3,000-5,000. in dollars 2010 to $600-800 in 2015 and in 2020 it is expected that their prices will fall to $80-120. Such a dramatic change in prices makes it very difficult to predict market growth, since:

• Declining prices cause rapid and unpredictable growth of the entire video surveillance market, including video analytics systems;
• The forecast of the market size becomes difficult, since video cameras occupy a significant market share, and forecasts, as a rule, poorly account for price changes.

Obstacles to market growth

The main limitation of the growth of the video analytics market is privacy considerations. This forces vendors to sell video analytics software bundled with other features to meet regulatory restrictions. For example, the European regulation on the protection of personal data GDPR (General Data Protection Regulation), which entered into force in May 2018, contains policies to limit video surveillance in public places and significantly limits the activities of many vendors.

^[3] from KiwiSecurity is an example of customization of video surveillance systems to comply with their GDPR standards.

Fig. 4 - 1. Example of how Privacy Protector works (source: privacy-protector.eu).

Video analytics vendors are forced to insert such applications into their platforms to comply with GDPR standards. Moreover, in most video analytics platforms, it is forbidden to track people's actions or somehow associate any actions with specific persons. The video data collected must be anonymized. In this regard, there are problems with the use of identification algorithms in video analytics applications. These privacy measures are supported by the governments of many countries of the world, which causes the need to inform people about the use and control of their personal data.

Growth opportunities

Applications for face recognition are always in high demand, including in personal devices, for the purpose of ensuring the security of devices. The face recognition function allows you to draw biometric identification identities by comparing facial features from the video stream with faces stored^[4]? database.]

Fig. 4 - 2. Biometric facial recognition (source: Starlinkindia).

With growing demand for both security applications and other facial recognition applications, these applications remain one of the main drivers of growth in the video analytics market. Various vertical markets such as retail, mass media and entertainment use video surveillance systems with facial recognition features.

Growth challenges

Intelligent video analytics systems are designed to minimize human participation in video surveillance, and to achieve other advantages, however, a number of problems arise when using them.

Insufficient infrastructure development in many developing countries (street lighting, signage, traffic lights, etc.) limits the effectiveness of the use of video analytics systems. False positives of analytical systems result in high costs for law enforcement agencies to find out whether an incident occurred in reality, as well as doubts about the reliability of video analytics systems.

In densely populated countries, such as China or India, public infrastructure is overburdened, and video surveillance requirements are increasing as the population grows. However, video analytics systems in such conditions are most susceptible to failures, even if they worked successfully in similar situations in other countries earlier.

Market Growth Forecasts

Various analytics companies indicate different growth figures for the video analytics market. This is due to the fact that the market is still unsettled, far from saturated, and very diverse. Different researchers attribute different spectra of applications and services to video analytics, which affects the original definitions of terms and industries of vertical markets.

According to the forecast of Inkwood Research^[5], the video analytics market will grow from $3647 million in 2018 to $20,450 million by 2027, with a CAGR CAGR of 20.61%. The market will grow the most in North America: nearly 80%. Europe will show growth of about 50%, slightly less growth - countries of Southeast Asia. The growth of the video analytics systems market in the rest of the world (RoW) will not exceed 20% (see figure below).

The largest growth in the video analytics systems market for 2017-2018 will be observed in retail (retail) - more than 50%, in transport - about 40%, in production - more than 20%, and in healthcare - about 30%.

The main vendors of the video analytics systems market, according to Inkwood Research: Cisco, Honeywell, IBM and Axis.

Fig. 4 - 3. Video Analytics Systems Market Growth Forecast (Source: Inkwood Research, 2018).

According to the research firm Allied Market Research, the size of the video analytics systems and services market in the world in 2023 will amount to more than $13.38 billion with a CAGR CAGR growth rate of 25.7%^[6].

McKinsey^[7] predicts the growth of video analytics applications with a CAGR CAGR of more than 50% over the next five years, mainly thanks to Internet Things applications. IoT Areas such as self-driving cars, automated factories and smart cities, as well as remote health care ones, will actively use various video analytics applications.

Fortune Business Insights forecasts the video analytics market to grow from $2488.5 million in 2018 ($835.9 million in North America) to $11965.6 million in 2026, with a CAGR of 22.67%.

Fig. 4 - 4. Video Analytics Market Growth from 2018 to 2026 (Fortune Business Insights Source).

According to Statista, revenue from hardware, software and video analytics services will reach $3 billion by 2022 with a CAGR growth rate of 19.6%^[8] Hardware^[9]. Market shares by industry are shown in the figure below.

Fig. 4 - 5. Video Analytics Services Market Sectors Growth Forecast (Source: Statista)

Overview of major vendors

Various analytical companies specify different lists of Top5, Top10, Top 20, etc. video analytics vendors, depending on their preferences, as well as areas of vendor products: security, face recognition marketing , in retail to trade , etc.

The list of more than 100 companies producing video analytics products, such as detecting and counting people, recognizing faces, vehicles, etc., in various scenarios is given below. The text contains hyperlinks to websites.

• 3VR, USA (Acquired by Idev)
• A.I. Tech, Italy
• Accenture, USA
• ACIC, Belgium
• ACTi, Taiwan
• Actuate, USA (formerly Aegis AI) a startup for recognizing types of weapons in video
• Alarm.com, USA
• Arcules, United States
• Agent VI, Israel
• Aimetis, Canada (acquired by Senstar)
• Aitek, Italy
• AllGoVision, India
• Anyvision, Israel
• Avidbeam, United States
• Avigilon, USA (acquired by Motorola)
• Axis, Sweden
• Axxon, Russia
• Boulder AI, USA
• Bosch, Germany
• BrainChip, USA
• Briefcam, Israel (Acquired by Canon)
• Calipsa, United Kingdom
• Camio, USA
• Checkvideo, USA
• Citilog, France (acquired by Axis in 2016)
• Cognimatics, Sweden (acquired by Axis in 2016)
• CyberExtruder, USA
• Davantis, Spain
• Dahua, China
• Deep Data, Y. Africa
• Deep Glint, China
• Deep North, United States
• Deep Science, USA.
• Deep Sentinel, USA
• Detec, Norway
• Digital Barriers, UK
• Disney Research, USA
• Emza, Israel
• Macroscop Eocortex (Macroscop), Russia
• EtherSec, UK
• Flame Analytics, Spain
• FLIR, US (formerly DvTel/IOImage)
• Foxstream, France
• Geutebrück, Germany
• GiantGray, USA, now "Omni AI"
• Gorilla Technology, Taiwan
• MangoDSP, Israel
• Hanwha, Y. Korea
• Hitachi Vantara, USA
• Hikvision, China
• Honeywell, USA
• IBM, USA
• i2vsys, India
• iCetana, Australia
• Intellivix, Yu. Korea
• Image Intelligence, Australia
• Impower Technologies, China
• IndigoVision, UK
• InPixal, France
• Instek Digital, Taiwan
• IntelliVision, USA (acquired by Nortek)
• intuVision, USA
• iOmniscient, Australia
• IPS, Germany
• Ipsotek, United Kingdom
• IronYun, Taiwan
• ISS, USA
• Jemez Technology, USA
• Kiwi Security, Austria
• Kogniz, USA
• Kognat, Australia
• Lighthouse, United States
• Magenta, Italy
• Megvii, China
• Microsoft, USA
• Mindtree, India
• Nanit, USA
• NEC, Japan
• Nest, USA.
• Netavis, Germany
• Netcamara, Argentina
• nFlux, USA
• Ngaro, United Kingdom
• NUUO, Taiwan
• Objectvideo, USA
• OpenALPR, USA
• OWAL, USA
• Paravision, USA
• Parking Spotter, Romania
• Pegasus, Brazil
• Prism Skylabs, USA
• Qognify, Israel
• PureTech Systems, USA
• RealNetworks, USA
• RetailFlux, Turkey
• Rhombus, USA
• RIVA, Germany
• Samsung SDS, Korea
• Scylla, United States
• SenseTime, Hong Kong
• Sighthound, USA
• Sightlogix, USA
• Smartervision, Netherlands
• Sony, Japan
• Sprinx Technologies, Italy
• Standard Cognition, USA
• TechnoAware, Italy
• TOP-KEY, Poland
• Umbo CV, Taiwan
• Uncanny Vision, India
• Uniview, China
• Vaion, United Kingdom
• VCA, UK
• Verint, USA
• Via: sys, Germany
• Videonetics, India
• Viisights, Israel
• Viion Systems Inc., Canada
• Vintra, USA, Spain
• Viseum, United Kingdom
• Visual Tools, Spain
• Vocord, Russia
• Zensors, USA.

There are quite a few vendors of video surveillance systems and video analytics in the world. In many sources, you can find lists in which, along with part of the above, there will be other companies. The video analytics industry is developing very rapidly, and new startups are constantly emerging in this area.

Often, various sources cite different lists of "TorXX" vendors of video analytics systems, in which various companies are present. This is because, as a rule, such vendors do not engage exclusively in video analytics, and have a fairly wide range of solutions. In addition, many video analytics solutions are "end-to-end," and can be used, for example, for analyzing big data and for other purposes.

The largest companies offering video analytics products, according to the Quora portal, are IBM, Bosch, Honeywell, Avigilon (acquired by Motorola in 2018). Large companies also need to include IntelliVision (Nortek), which has already posted more than 4 million CCTV cameras in the world^[10]

According to Business Wire^[11], the following companies are the largest vendors of systems and video analytics software:

• Agent Video Intelligence develops software solutions for video analytics based on an open architecture, as well as solutions for security, security and business intelligence applications. The company's products can be supplied both in the "boxed" version and in the cloud.
• Bosch Security Systems designs, develops and supplies a wide range of security devices for various applications, such as video surveillance, access control, communications, and highly specialized intrusion and ignition detection devices.
• IBM offers various types of equipment, software, infrastructure consulting and maintenance services, including video surveillance and video analytics, for various industries such as the automotive industry, healthcare, media and entertainment, mining, retail.
• IntelliVision develops various solutions for intelligent video analytics, smart cameras, and automated monitoring.
• Honeywell International develops a solution and manufactures system products for automation of production and safety for various industries and regions of the world.

Market segments

The video analytics market can be segmented based on the following types:

• components;
• applications;
• deployment models;
• vertical industries;
• geographical distribution.

Video analytics software is an organic part of intelligent video surveillance systems. The video analytics software market size can be from 3 to 10% of the total video surveillance market, depending on the functionality and analytics requirements.

By component type, the global market is segmented into

• software sector
• service sector.

It is expected that the software sector will account for the main market share for the near term until 2025^[12].

By application type, the video analytics market is segmented into:

• crowd detection & management;
• facial recognition & detection;
• perimeter intrusion detection
• incident detection
• Traffic & parking management
• other special applications.

Incident recognition applications are the main type of video analytics applications required for various video surveillance applications and will have the largest market share for the near term.

By deployment model, the video analytics market is segmented into

• video analytics cloud services market;
• systems deployed at the user site (on-premium).

For the period up to 2025, the on-premium regime is expected to have the largest market share. By Vertical Industry, the market is segmented into

• Transport;
• Smart City;
• Critical infrastructure;
• Corporate sector;
• Trading;
• Others.

Vertical industries

Transport

The demand for intelligent video surveillance systems in transport is constantly growing at a faster pace than other video analytics markets. They are necessary to improve public safety, prevent and prevent offending by anti-social elements, reduce the danger of terrorism, and increase public responsibility and involvement in incident investigations.

Fig. 4 - 6. Video analytics applications for transport (source: DTI).

Intelligent video surveillance systems provide various benefits for passengers, carriers and transport agencies including:

• Ensuring public responsibility, including confirmation and evidence of individual complaints and requirements in order to strengthen work on valid complaints and reject false complaints;
• Suppression of undesirable, antisocial and criminal behavior of individual individuals, increasing the public prestige of public transport and improving the security of users;
• Ensuring the safety of drivers and passengers;
• Protecting property and reducing losses from vandalism, graffiti and property damage to transport infrastructure and passengers;
• Reduced time and improved response to incidents;
• Improving the quality of passenger service in transport, which translates into an increase in the profitability of transport infrastructure.

The size of the global video surveillance market (including cameras, network and computing infrastructure, storage systems and software) on various types of transport infrastructure is shown in the figure below.

Fig. 4 - 7. Global Transportation Video Surveillance Market Size (Source: Market Research Future, November 2019)

Urban Infrastructure Management

According to Cisco, the volume of the urban infrastructure management market (including intelligent video surveillance systems) can have a volume of services and received value from ownership, TVO^[13] (Total value of ownership), up to $66 billion.

Fig. 4 - 8. Smart City Digital Systems Market Assessment (Source: Cisco, 2019).

The urban video analytics systems market can account for 7-10% of the total global video analytics systems market with an annual growth rate of 25-30%.

Critical infrastructures

Intelligent video surveillance systems, in particular, using artificial intelligence, are very important for organizations operating critical infrastructure systems: generation and distribution of electricity, water supply, gas supply.

For them, integrated systems are very important using advanced video analytics and facial recognition methods, which are able to automatically recognize potential threats and important events in a short time (almost in real time), as well as promptly send notifications to the relevant personnel, departments and agencies.

Electrical substations, which usually require increased safety measures, are particularly susceptible to such threats due to their role in the energy distribution process, as well as the nature of the equipment used in them, the bòlshaya part of which poses a danger to human life.

Manufacturers of video and analytics systems are developing solutions for the critical infrastructure market as one of the most important. For example, the Chinese company Dahua offers explosion-proof cameras (IP68 protection indicator), which use high-quality optics for reliable protection. Such cameras are used for surveillance and analytics in aggressive and corrosive environments^[14], for example, in chemical plants and in the oil and gas sector.

The figure below shows the regions of the world that have the highest growth rates of the video analytics market for the critical infrastructure of the economy (green), moderate growth (yellow) and low growth rates in this sector (red). For Russia, the source data is missing.

Fig. 4 - 9. Video Analytics Market Growth Themes for Critical Economic Infrastructure (Source Mordor Intelligence, 2018).

^[15] security infrastructure and security solutions market^[16]which intelligent video surveillance and video analytics systems are a part, is shown in the figure below.

In 2017, the market for critical infrastructure protection facilities was estimated at $110 billion. By 2022, it is expected to grow to $153 billion with an average annual growth rate of 6.8%.

Fig. 4 - 10. Growth of the market for means and solutions of protection for critical infrastructure (source: ©2019 MarketsandMarkets Research).

The growth of the market is due to the tightening of government regulation of the security market and the growing number of incidents and cyber threats aimed at critical infrastructure enterprises. Critical infrastructure security is expected to actively use technologies such as the Internet of Things (IoT), big data analytics and video analytics in cloud systems.

This market segment will be most active in the countries of Southeast Asia and the Middle/Middle East.

Mordor Intelligence gives more modest estimates of the critical infrastructure protection market, which is explained by a different approach to classifying services and functions.

Fig. 4 - 11. Critical Infrastructure Protection Market Growth 2018-2024 (Source: Mordor Intelligence).

The global video analytics software market for critical infrastructure is estimated at $250-300 million by 2023 with an annual growth rate of 20-25%.

Corporate sector

In the digital economy, based on the analysis of data and the extraction of useful insights from the collected data, including video data (not only from video surveillance cameras, but also, for example, from corporate video conferencing, webinars, training videos, etc.), video analytics is one of the main means for developing a strategy for the development of the enterprise. Huge amounts of video data accumulate at enterprises, and therefore it is very important to have the appropriate analytical tools to extract the necessary^[17] information from them].

Business leaders are looking to improve employee communications and work environments, so they often wonder: how video can improve a company's business performance. For example, they may be interested in the following questions:

• Has the quality of video conferencing improved compared to the previous reporting period?
• Has staff efficiency improved when using personal video viewing devices?
• Which parts of the enterprise most fall into the field of view of video cameras?

In order to get answers to these questions, good video analytics tools are required, not limited to spreadsheets and means of counting people and time. Effective information visualization tools are required to make it visual, get an insight, and be available to all stakeholders in the business process.

Video analytics is used in many digital technologies of the Industry 4.0^[18], such as:

• Predictive maintenance of equipment and products;
• Production management systems;
• Comprehensive supply planning;
• Internet of Things IoT, of which the video cameras are a part;
• Digital product and production line twins;
• Robots assistants, smart robots, robotic processes;
• Artificial intelligence;
• Virtual and augmented reality AR/.VR

Fig. 4 - 12. Application of video analytics in new technologies Industry 4.0 (source: PricewaterhouseCoopers, 2018).

Retail

Using the "heat map" video analytics function in the video surveillance system of a trading enterprise, you can get a lot of useful information, for example, about the number of buyers lingering near certain goods on shelves. This information can be used to more effectively market goods in demand, as well as promote goods that escape the attention of buyers. In addition, this helps to better plan retail space, for example, a decision may be made to organize closer passages to low-visited places of the trading floor so that goods located there are better sold.

Fig. 4 - 13. Store heat map (source: ACTi).

The figure below shows an example of the NEC Enhanced^[19] for^[20] retail user interface].

Fig. 4 - 14. An example of a video analytics system user interface for retail (source: NEC, 2018).

According to Tractica, Retail is the largest sector of video analytics services and systems, with a volume of approximately $800 million in 2019 and an increase to $1.1 billion in 2022 with a CAGR of about 20%.

Fig. 4 - 15. Retail Video Analytics Market Growth (Source: Tractica).

Market dynamics

Various analytics companies in their research show different growth dynamics of the video analytics market. For example, Mordor Intelligence shows an average annual market growth of 28.89%^[21], while other companies are more modest in their estimates.

Fig. 4 - 16. Optimistic Growth Rate Forecast for Video Analytics Market (Source: Mordor Intelligence).

Technavio predicts an even higher growth rate in the video analytics market from 2016 to 2021 - up to 31%^[22].

Fig. 4 - 17. Market dynamics forecast from Technavio.

According to Technavio, the market is highly fragmented, and many players will work on it. About 38% of the market growth will be, according to Technavio, in the American continent. Although other analysts believe that the main growth will occur in the Asia-Pacific region, and mainly in China.

So, for example, the Markets and Markets company (see the drawing below) considers that the video surveillance market (which part in 4-10% is the market of video analytics) will be in the lead in the Pacific Rim^[23].

Fig. 4 - 18. Video surveillance market size in 2023 (source: Markets and Markets, TAdviser, 2019).

The main growth sector of the market, according to Technavio, will be smart city systems, although many other analysts believe it will be retail.

According to TAdviser analysts, the following factors may be the cause of such unequal estimates:

• The technological basis of video analytics systems is undergoing large and rapid changes:
  • Use of big data analysis systems;
  • Growth of resolution and other process parameters of video cameras;
  • Use of neural networks (artificial intelligence) to analyze video images;
  • Increased bandwidth of communication networks;
  • The development of new network architectures, in particular Edge Computing and the placement of analytical systems at the border of the network and directly in user devices;
  • Growing video storage capabilities, including increasing and reducing the price of SSDs.

• Uncertainty ^[24]. video analytics systems and markets:
  • Complexity of classification of applications;
  • Uncertainty of market sectors: some companies may refer to video analytics, for example, BI (Business Intelligence) systems, others distinguish them into a separate market sector, etc. Also, for example, there are discrepancies regarding Smart City systems, which, according to some analysts, will be the main market growth sector. Others believe that the main growth sector will be Retail, which can be classified as a Smart City system and considered a separate sector.
  • Some companies may assume that only software platforms belong to the video analytics market, others also take into account the equipment market, including servers and video cameras with video analytics functions;
  • Lack of consensus on whether video analytics systems can include Video Surveillance as SaaS a Service (V) cloud services, which can also provide analytical functions.

However, various market indicators show positive growth dynamics in the use of video analytics systems and services and horizontal growth of the market, i.e. the involvement of those who were previously either skeptical about the capabilities of video analytics or were simply not aware of these capabilities.

For example, a survey of subscribers to SDM Magazine^[25] shows that in 2019, interest in investments in video analytics in enterprises is growing significantly compared to previous years.

Fig. 4 - 19. Growth expectations for investment in video analytics (source: SDM Industry Forecast Study, 2019).

The survey results show a 13% increase in interest in video analytics systems and services in 2019 compared to 2018. At the same time, the number of people who believe that investments are not growing is decreasing.

The dynamics of the video surveillance systems market, which mainly uses video analytics, is shown in the forecast of the information portal Statista^[26] By 2023, the video surveillance market is expected to be $62-63 billion.

Fig. 4 - 20. Video Surveillance Market Growth Dynamics (Source: statista.com).

Market Sector Interference and Synergies (End-to-End Technologies)

The factors of ambiguous estimates of the video analytics market by various analytical companies should also include the mutual influence and synergy of the market sectors, the reuse of "end-to-end," which causes ambiguous estimates in forecasting.

With the development of various digital technologies, the so-called "end-to-end digital technologies" (SCT) are becoming more and more important. These are technologies that can be used without special adjustments and modifications in various market segments.

This approach allows us to achieve high efficiency of new developments, as well as achieve a synergistic effect, which allows, by introducing a particular technology in one market segment, to significantly accelerate the development of other segments.

Such SCTs are defined in the Program for the Development of the Digital Economy of Russia ^[27]].

In particular, the SCT "Neurotechnologies and Artificial Intelligence" indicates the following types of subtechnologies, which include the functions of video analytics:

• computer vision;
• promising methods and technologies in AI;
• recommendation systems and intelligent decision support systems.

Fig. 4 - 21. Sub-technologies of the Program for the Development of the Digital Economy of Russia, including the functions of video analytics (source TAdviser, 2019).

There are three segments of video analytics with end-to-end technologies that are most widely used in various industry markets.

• Video analytics of non-standard behavior
• Automatic license plate recognition
• Facial Recognition Analytics

In all three segments, the common technology is neural networks, which serve as a common digital platform for creating the functions of the three listed segments.

Video analytics of non-standard behavior

Analysis of non-standard behavior is based on deep learning (Deep Learning) of neural networks as part of artificial intelligence AI (AI, Artificial Intelligence) ^[28]

Trained on a variety of scenes, the neural network recognizes the poses and style of behavior (walking, running, standing still for a certain time, etc.) of people in the field of view of the video camera. In this case, a large array of data is formed, containing information about the standard behavior of people. As soon as a person in the frame begins to behave in a non-standard way, for example, goes against the flow of other people, waves his arms, aimlessly "loitering," or performs other unusual actions, the AI algorithm issues a warning to the operator of the video surveillance system and highlights such a person with a frame on the monitor screen.

In addition to automatic training of the neural network, you can also create software detectors of certain positions and movements, which are developed according to the requirements of certain customers (state or commercial). At the same time, additional training of the neural network will not be required.

Neural networks and AI are applicable in all segments of the video surveillance market to extract information about the observed scene and the objects present on it from the image stream.

Retail

In Retail to trade , neural network-based video analytics of non-standard behavior are used to analyze the work of store staff and the behavior of customers. In the process of such analysis, it is done to identify the most visited places ("heat card"), recognize the faces of individual customers (regular buyers, buyers who make purchases for large amounts, and, possibly, suspicious persons) control the filling of shelves and the correct arrangement of goods, and identify potentially fraudulent transactions of cashiers.

It is also possible to control the response time of consultants to the appearance of a visitor. All these tasks are aimed at improving the efficiency of the trading enterprise.

The figure below shows a screenshot of a video analytics system in a shopping center. The frame recognizes non-standard behavior (Loitering), Slip & Fall, Object Detection in a Crowd, Graffiti Detection, and Facial Detection, which is on some special list (suspicious persons, or vice versa, important customers).

Рис. 4 - 22. Пример видеоаналитики нестандартного поведения в retail (source: https://www.nvscanada.ca/).

Integration of video surveillance systems with POS terminals

with System integration video surveillances POS terminals it serves to improve business efficiency by synergy of two processes: calculation for goods at the cash register and control using a video surveillance system. This combination of two functions serves to fully analyze incoming data from the POS terminal along with algorithms for determining suspicious behavior.

For example, an intruder cashier purchases a discount card and uses it in settlements with buyers who do not have this card, thereby providing them with an illegitimate discount, but accumulates bonus points for himself. The discount reduces the profit of the store owner and, in fact, such an unscrupulous employee steals his employer.

The non-standard behavior video analytics system recognizes too often the application of a bonus card to the terminal at the checkout, and notifies the relevant personnel. The synergistic work of such functions gives new areas of application of video analytics, both in retail and in other areas.

The share of fraud in various industries in the United States based on 400 thousand cases of the insurance company Hiscox is shown in the figure below^[29].

Fig. 4 - 23. Share of fraudulent activities in various sectors of the US economy (source: CBS MONEYWATCH, 2017).

The average amount of losses from one fact of fraud is shown in the figure below, from which it is clear that the funds invested in video analytics pay off quickly enough.

Fig. 4 - 24. The average amount of losses from one fact of fraud in various sectors of the US economy (source: CBS MONEYWATCH, 2017).

One of the main economic effects of video analytics is to reduce the damage from fraud and theft. A study by insurance company Hiscox showed that in the United States, the average damage from fraudulent actions of one unscrupulous employee over five years is $2.2 million.^[30]

Experts from the analytical company Lexis Nexis Risk Solutions calculated that for every dollar stolen from the business there is $2.66 spent on the costs associated with the theft of one dollar (along with the replacement of the lost product)^[31].

Transport

In the transport segment, neural networks are used to recognize a non-standard driving style (dangerous driving or traffic violations), statistical analysis of traffic flows, counting the number of passengers to control fare, to control the time and correctness of parking.

Automatic license plate recognition

Automatic License Plate Recognition (ALPR) is a video analytics function designed to control traffic violations and maintain parking lots. So, for example, ALPR allows you to track the number of empty parking spaces, this is more efficient than ultrasonic and magnetic sensors.

ALPR can be integrated with payment systems: parking time is fixed by ALPR; withdrawal of funds for parking takes place automatically from an attached bank card or other means of payment. ALPR improves the quality of customer service: for example, the system reads the number of the approaching car, and, if it is registered, immediately opens the barrier.

Facial recognition

Facial recognition in video analytics is used in two modes: simply identifying a person's face and recognizing facial features to determine a person's identity.

• Definition of Person
  The definition of person is used, for example, to trade in retail to determine the number of buyers passing through a certain border, for example, crossing the border of the entrance to the trading floor, or delaying at a certain trading stand. At the same time, it is possible to track the gaze to determine which products are most looked at by buyers, and what emotions their faces express. The route of a particular person in the observation area can be tracked by face. Advanced video analytics systems are able to determine the emotions that a customer who looks at a particular product can express: surprise, admiration, skepticism, disgust, etc. Quantitative analysis of such data helps to make adequate marketing decisions by the management of the trading enterprise.
• Identity Definition
  The definition of a person's identity means unambiguous authentication of a person, with partial use of personal data. To do this, the person must be registered in a database. Systems are being developed for the analysis of emotions by facial expressions, and for the analysis of interest in the direction of gaze.

All these functions were implemented thanks to neural networks with deep learning^[32].

Fig. 4 - 25. An example of a neural network with deep learning for face recognition (source: ICS Media, 2019).

See also

• Video Analytics Systems - Systems and Project Catalog
• Video Surveillance Systems - Systems and Project Catalog
• Information Security - Biometric Identification - Catalog of Systems and Projects
• VisionLabs - VisionLabs Luna
• Sinesis Cepheus Video Analytics Complex - Synes
• Synes: Video analytic device - Synes
• Kipod Cluster - Synes
• CasRetail is Synes
• VisorJet with video analytics for rotary camera - Elvis-NeoTech
• ELISE Semantic Processor (ELVEES Image Semantic Engine) - Elvis-NeoTech
• Video Intellect: Video Analytics System - Video Intellect
• Video Synopsis - BriefCam
• R-Style: PRIME Video Analytics - R-Style Computers
• Business Analytics: Video Analytics for Shopping Centers - Business Analytics
• How computer vision systems are changing logistics and medicine
• Vocord
• Yandex Data Factory
• Cognitive Technologies
• Macroscop (Satellite Innovation)
• Aggregator
• Integra-with
• 3iTech (formerly 3i Technologies)
• LogistiX (Logistic-Tech)
• Intelligence Retail
• Wireless Analytics (WiFi Analytics)
• Wireless analytics in Russian retail
• Visitor counting technologies
• Retailbase: WiFi Analytics - NPO Analitika

Notes