MarketsandMarkets: forecasts for the development of the video analytics market 2017-2022

The video analytics market will grow from 2.61 billion US dollars in 2017 up to 11.17 billion US dollars by 2022, with total annual growth rate of 33.7%. According to experts, the key factors influencing the video analytics market are: high demand for effective video surveillance and government investment in new technologies to improve public safety infrastructure, and, consequently, demand for effective algorithms for processing a growing amount of data.

Facial recognition drives video analytics market growth, with increasing acceptance of this technology as one of the most advanced security tools by crime investigation and law enforcement agencies around the world. Video analytics based on , is a powerful application that can be easily deployed to improve your security posture.

In addition, video analytics is increasingly being used for non-security applications, such as in retail, where facial recognition is used to identify loyal customers and give them preferential treatment, resulting in high levels of loyalty.

Experts predict that the most effective business model would be to connect video analytics on demand.

Concerning geographic segmentation, the North American region is the leader, but experts predict active growth for the Asia-Pacific region due to initiatives.

Major video analytics providers experts name: Cisco Systems (USA), Honeywell International Inc. (USA), IBM Corporation (USA), Axis Communication (USA), Video Intelligent Agent (USA), AllGoVision Technologies Pvt. (India), Avigilon Corporation (Australia), Aventura Technologies Inc. (USA), Genetec Inc. (Canada), I2V System Private Ltd. (India), Intellivision Technologies (USA), IntuVision (USA), PureTech Systems Inc (USA) and Qognify (USA).

Video analytics technologies

The impact of video quality on analytics capabilities

As the resolution of surveillance cameras increases, the capabilities and functionality of video analytics expand. Companies offering video analytics solutions primarily rely on modern models megapixel video cameras. However, even private companies do not always have a sufficient budget to quickly modernize the system. As for government projects, in addition to the allocation of funds, it is necessary to overcome bureaucratic barriers from regulations and approvals.

Therefore, some manufacturers develop software for analyzing video, even if it is not of very good quality.

"Our algorithms even work with video streams that have low frequency frames or low resolution, which is traditional for fragments that are most often used in criminal investigations. We provide full package reporting documentation, including annotation, editorial, explanation and storyboard", says Sarah Doyle, Managing Director of Kinesense.

Using video analytics based on cloud platforms

At the same time, the search for an alarming event is often carried out in the archive rather than in real time.

At the same time, from an algorithmic point of view, analyzing a video archive is simpler than analyzing a situation in real time. When something has happened, the signs of the event whose record needs to be found are already known. For example, this could be a description of eyewitnesses, a sketch of a suspect, or a car license plate number. Therefore, there is no need to configure detectors in advance for unknown alarm situations. It is enough to analyze the video at the time of recording, extract raw data from it that can be searched, and save it. This data includes biometric characteristics faces, recognized license plates and metadata describing the movement of objects in the frame, their coordinates at different times, sizes, colors.
The more information is extracted from the video stream during recording, the more search opportunities there will be in the future.
Fragment of the TK article: .

Video surveillance as a service enables a new analytics business model: a one-time request to examine video to detect any events. The customer uploads the video archive to the cloud server, where the supplier’s tools search and analyze events, and generate a report.

Automatic License Plate Recognition (ALPR)

Automatic license plate recognition or Automatic License Plate Recognition (ALPR) is a video analytics function that makes it convenient to manage traffic flows and maintain parking lots. For example, ALPR allows you to monitor the number of empty parking spaces, which is more cost-effective than using ultrasonic and magnetic sensors.

In addition, automatic license plate recognition can be used when integrating with payment systems: parking time is recorded by ALPR; write-off Money for parking occurs in automatic mode from an attached electronic wallet topped up by the user.

Another function that increases ease of use and response speed is the ability to pre-register for events: the system analyzes information about the license plate number of the approaching car, and, if it is in the database, immediately opens the barrier.

Video analytics using artificial intelligence

Deep learning mechanisms based on artificial intelligence make it possible to improve video analysis mechanisms, moving from analyzing events that have already occurred to preventing them.

The development of video analysis algorithms has remained one of the main trends in video surveillance for many years. And considerable progress has been made here, especially in highly specialized areas such as facial recognition and license plates. Thanks to the use of neural networks and machine learning, a computer is almost as good as a human in solving these problems, and in certain cases even surpasses it. But in solving non-specific problems, a computer is still far from a human. Video surveillance systems have not yet learned how to reliably detect inappropriate or potentially dangerous behavior. Certainly, neural networks Here too they look promising, but there is a long way to go before the practical implementation of “a system that assesses the situation like a person does.”
Fragment of the article TK: Searching in video archives will make video surveillance much more effective.

Face recognition analytics

Panasonic introduced new I-Pro SmartHD video surveillance cameras with advanced business intelligence functions. The analytical face recognition platform for use as part of the WV-ASM200 and WV-ASC970/WV-ASM970 software not only allows for real-time face recognition, but also maintains statistics of those caught in the camera's field of view, taking into account their gender and age, which also are recognized by the system.

In the case of searching for persons in a database, the result can be presented in chronological order, which allows you to analyze the movements of people.

As a rule, a video surveillance system that supports facial recognition analytics is built on cameras that can adapt to constantly changing environmental conditions: rain, snow, fog, changing lighting conditions - without compromising the quality of facial recognition. In addition, the standard camera software includes features such as automatic tracking and scene change detection: cameras can detect abnormal situations and analyze suspicious movement without operator control and immediately alert the control center.

All devices can be remotely controlled via smartphones or tablets using special application Panasonic.

Articles

Select year: Select month:

• Imou is a new brand of smart video surveillance
  

  The slogan of the new brand: “IMOU - live wisely!”, which is reflected in the concept of easy installation, as well as the presence of basic video analytics functions

• Facial recognition in a video surveillance system: why and how it works
  

  There are two main ways to implement facial recognition in video surveillance. And each of them has its own advantages and disadvantages.

• Think Value, Not Megapixels: How to Sell HD Security Cameras
  

  More a high resolution in itself, of course, is great, but what really matters for integrators and users of security systems is that when implemented, it allows you to provide a number of useful and interesting functions.

• Axis network technologies for surveillance and control
  

  The comprehensive Axis system covers all key aspects of industrial plant security: protection against intrusion, ensuring the efficiency of production processes, compliance with occupational safety and health regulations.

• Video surveillance system for a hardware store based on Synology
  

  Following these repeated thefts, the building supply store decided to look for a cost-effective, centralized and flexible security solution.

• High speeds and safety: Bosch equips the Misano race track in Italy
  

  The Misano International Circuit was opened in 1972 near the town of Misano Adriatico in northern Italy and has become one of the most famous motorcycle racing tracks in the world.

• What you need to know about video analytics before implementing it
  

  If the video analytics system can run on a camera, it will likely require high-end camera models, since these cameras are equipped with the most powerful processors.

• Behavioral analytics in video surveillance
  

  Artificial intelligence technologies have made possible solution extremely difficult tasks. One of these tasks is the analysis of human behavior from images obtained using a CCTV camera. This new type of video analytics can be used to great effect in the public safety and commercial sectors.

• AXIS Perimeter Defender - scalable perimeter protection tool
  

  Axis has developed a highly reliable, scalable perimeter protection solution – AXIS Perimeter Defender. The video analytics application, combined with Axis network cameras, creates a highly effective system for automatically detecting people and vehicles.

• A New Era of Smart Video Surveillance and Security
  

  With the advent of IP cameras and cloud-enabled technologies, the video surveillance industry has changed rapidly. And yet there are even more potential opportunities.

• Analytical capabilities of DVRs in analog video surveillance systems
  

  If the goals and objectives of the video surveillance system are clear, then the correct
  selection of equipment will significantly expand its capabilities.
  In particular, we are talking about the analytical functionality of professional
  DVRs.

• Pelco ushers in a new era of intelligent video surveillance and security with technology alliance with IBM
  

  When it comes to video surveillance tools and services, the industry has barely scratched the surface of what is possible today. This does not mean that progress has not been made to bring innovative cameras, IP video technologies, monitoring and detection systems to the market, but we can already say that the next important stage of development is on the horizon.

• Thermal imaging video analytics
  

  Thermal video analytics help operators effectively respond to threats by providing the right information at the right time. But this is subject to proper configuration of the thermal imager.

• Intelligent people counting function to optimize CCTV in stores
  

  Axis CCTV systems with people counting support enable efficient collection and analysis of customer data.

• Confidence in video analytics is being restored and demand for these technologies is growing
  

  After a rocky start, improvements and more modern applications are restoring confidence in video analytics and increasing demand for the technology. When it comes to new advances in video analytics, everyone is interested not just in describing the technology, but in how it has changed and how it can be used in innovative applications.

• IP video surveillance and analytics
  

  In fact, such a panel is not just a call initiation device as part of an intercom system, but also a full-fledged IP camera as an additional monitoring and video recording tool for obtaining subsequent analytics and statistics.

• Is video analytics relevant today?
  

  Since their first introduction, intelligent video analytics systems have gone through stages of premature false heyday in terms of justified implementation of these solutions as a reliable and cost-effective addition to traditional elements of a TV surveillance system. Despite the fact that video analytics was predicted to have a promising future, the expected widespread implementation did not happen for a number of objective reasons.

• Integration in security systems and the place of access control systems in them
  

  Will the dominant place of access control systems (ACS) in integrated security systems (ISO) change? Will real IP technologies appear in ISO, and not just the use of Ethernet networks as communication channels instead of exchanging data over RS ​​232 or RS 485 interfaces? What are open protocols, and how to achieve them?

News

Select year: Select month:

A conference was held in Moscow dedicated to the emergence of a new brand. At the opening ceremony the General Director Dahua Technology Rus Zhao Shenbo emphasized: “IMOU is an IoT ecosystem that combines a cloud solution, smart devices and intelligent technologies. The brand’s ideology demonstrates how high-quality video surveillance and advanced technologies ensure the safety of home and business.".




Traditional airport video surveillance systems contain a large number of video cameras, and the monitoring service staff is quite large. Maintaining such systems requires a lot of human resources, which means that they extremely expensive to operate.




Company Axis Communications expanded the production line at the plant Jabil in Tver, one of Axis' production partners in Europe, due to high demand for the company's products. The lineup, presented at Russian production, has been replenished with cameras with new generation chipARTPEC-7, with elements of artificial intelligence. At the moment, Axis's own production facilities produce 16 models that are most in demand in the world. Russian market cameras Thus, their number increased by 60% compared to 2016, when the company launched a local production line in Russia.

Company Dahua Technology, the world's leading provider of video-centric smart IoT solutions and services, introduced NVR 4.0 with a new user interface and a new interface that simplifies the user experience. New NVR 4.0 Supports major Dahua DVR models including NVR2000-4KS2, NVR4000-4KS2, NVR5000-4KS2, NVR4000-I, as well as NVR5000-I.

Company Axis Communications, a global leader in network video surveillance, presented a range of high-tech security solutions within Axis Partner Day 2019, which took place on October 22, 2019 in Moscow. In the event, the main themes of which were and cybersecurity, video analytics, specialized solutions for retail, transport and critical infrastructure, attended by industry experts, Axis specialists, as well as the company’s partners and customers.




IDIS, the largest in South Korea manufacturer of complex video surveillance systems invites you to take part in a short stream dedicated to IDIS Deep Learning Analytics (IDLA) video analytics and examples of practical cases.




Company Dahua Technolog y, the world's leading provider of smart IoT solutions and services, has released IVSS 2.0 (IVSS7000-I), updated series DVRsWith the latest algorithms artificial intelligence with deep learning. This series opens up new possibilities for smart security solutions.




Axis Communications announced the release of economical and AXIS P5655-E high-performance PTZ network camera. It allows high-quality shooting in four zoom scenarios, and is also equipped new generation chip ARTPEC 7, which increases the power of analytics and image compression.




October 10 in Yekaterinburg will take place multi-brand practical seminar for installers, designers, security managers, where representatives are top officials of manufacturing companies ProxWay, ZKTeco, Stelberry and NOVIcam will talk about the latest technological advances that can be successfully integrated into modern security systems.




In September 2019 the annual IDIS Global Partners Summit (GPS) specialists in the field of security and video surveillance, which was attended by delegates from 23 countries. GPS participants learned about the latest developments in the field A.I., and also got acquainted with IDIS roadmap for 2020.

In full voice!




Axis Partner Day 2019 is a significant event in the field of security and video surveillance, which traditionally brings together leading industry representatives and Axis specialists, as well as the company’s partners and customers. The event will be held under the slogan “ In full voice!” – and will consist not only of presentations. This year, the goal is to talk with participants, exchange opinions, review the stories of implementation of Axis solutions at various sites, talk about market trends and hear the opinions of third-party experts.




Box IP Video Camera Series IDIS replenished new model in modular design DC-V3213XJ with 4.0 mm lens. A 2MP camera is optimal for compact video surveillance. The video camera consists from a remote miniature video module and the main unit, which are interconnected by UTP cable up to 30 meters long.




Newest technologies biometrics in video surveillance and professional audio recording. We invite you to attend a practical seminar October 10 for installers, designers, security managers.




Meeting with top officials of manufacturing companies for installers, designers, security managers.




IDIS, the largest manufacturer of integrated video surveillance systems in South Korea, invites you to take part in the training seminar “How to implement a project using IDIS equipment.” The event will be of interest to technical specialists from integrator companies, installers and installation organizations.




IDIS, South Korea's largest manufacturer of integrated video surveillance systems, invites you to participate in webinar to review the IDIS line of panoramic IP video cameras and them practical application.




Axis Communications expanded its portfolio of video surveillance products and solutions with the launch of a line of discreet and compact cameras AXIS M42, led indicator AXIS I/O Indication LED, and Axis Live Privacy Shield technology, which ensures the privacy of people by applying masks to moving objects and has no analogues on the market. In addition to retail, these solutions can be used in healthcare, education, manufacturing and transportation facilities.

Axis Communications has expanded its explosion-proof camera portfolio with two new models: XF Q1785 and XF Q1645.




AgencyMemoori , a consulting company based in Stockholm that provides independent market research, business analytics and consulting on Smart Building technologies, has published analytical material revealing the specifics of the Russianmarket of security systems, video surveillance and fire alarm systems. We present to your attentionforeign experts' views on the situation .




Software Video Management System (VMS) plays a huge role in IP video surveillance systems, providing operators with effective control and a seamless experience.

IDIS introduces the new DV-2116 video analytics appliance, which increases the power of video surveillance systems with accurate deep learning analytics.




IDIS introduces a new high-speed PTZ IP video camera DC-S6283HRXL with a powerful 36x optical zoom, autofocus, Hall sensor and LightMaster technology for optimal image quality, which allows you to see the smallest details of objects located hundreds of meters from the camcorder itself.

Company Dahua Technology, the world's leading provider of IoT video surveillance solutions, has launched a series NVR4000-I with artificial intelligence, which included models DHI-NVR4208-8P-I, DHI-NVR4216-16P-I, DHI-NVR4216-I, DHI-NVR4416-16P-I, DHI-NVR4432-I and DHI-NVR4832-I.

From March 19 to 22, Moscow hosted the main industry event of the year - the International Exhibition Securika Moscow 2019 - the largest exhibition of technical security equipment and equipment for security and fire protection in Russia and the CIS countries.

Company Tatris Hanwha Techwin, introduces a new high-speed PTZ IP camera Wisenet XNP-6550RH with 55x zoom.




Axis Communications produces automobile access systems with automatic license plate recognition in Russia.

Company Tatris, as an official distributor Hanwha Techwin, introduces a new network box camera with 32x optical zoom Wisenet XNZ-6320.

September 20 company Bosch Security Systems organized a real holiday for designers and integrators Bosch Security Fest 2018, took place in the center of Moscow!




Bosch Security Systems presents a solution In-Store Analytics, with which retailers will be able to obtain valuable information about the movement of customers in the store, optimizing operations, attracting customers and increasing sales.

IP Security Forum, organized annually by ITV | AxxonSoft celebrates its first anniversary - 5 years! Every year the scale and scope of the event grew, and this year it goes beyond Russia for the first time.




IP Security Forum, organized annually by ITV | AxxonSoft celebrates its first anniversary - 5 years! Every year the scale and scope of the event grew, and this year it goes beyond Russia for the first time.

Company Tatris, as an official distributor Hanwha Techwin, is two new systems for small retail chains and enterprises, jointly developed by Hanwha Techwin and A.I Tech.




Company Dahua Technology, a leading global video surveillance solutions provider, has officially unveiled its first deep learning network video recorder, bringing artificial intelligence (AI) capabilities to a variety of applications.




Team Macroscop* took part with her own stand at the international exhibition IFSEC International 2018, which took place June 19-21 in London, UK.

Company Tatris, as a partner Panasonic presents new 360 degree dome 5 megapixel network cameras WV-S4150 And WV-S4550L.

Company IDIS at the exhibition ASIS 2017 in Dallas announced its own development in the field based on machine learning and artificial intelligence IDIS Deep Learning Analytics (IDLA) and outlined the prospects for deep video analytics in the industry.




Company Tatris, as an official distributor Hanwha Techwin, introduces a new analog camera for indoor installation Wisenet HCZ-6320P, made in a standard case and supporting work in two formats - AHD and CVBS. Availability of built-in zoom lens with maximum magnification 32x allows you to use the new product for video surveillance in a large space.




TD “Lider-SB” announces the availability of a new IP video camera Dahua DH-IPC-HDW4231EMP-AS-0360B-S2.

Russian developer of distributed information systems, intelligent video surveillance systems, biometric facial recognition and - Intra - became a sponsor of the 8th Intelligent Video 2.0 conference at the All-over-IP 2017 forum.

On October 5, Murat Altuev, President of ITV, will speak at the IP Forum in St. Petersburg | AxxonSoft. in the report "Axxon Next - roadmap" he will talk about how the innovative video surveillance system and the company’s key technologies will develop.




TD "Leader-SB" announces new capabilities of the event video monitoring device CHEKTV III HD.




This autumn ITV | For the fourth time, AxxonSoft will hold the annual IP forum, bringing together leading manufacturers, distributors, integrators, installers and customers of security systems.





This autumn ITV | For the fourth time, AxxonSoft will hold the annual IP forum, bringing together leading manufacturers, distributors, integrators, installers and customers of security systems.





This autumn ITV | For the fourth time, AxxonSoft will hold the annual IP forum, bringing together leading manufacturers, distributors, integrators, installers and customers of security systems.

• 
  

YouTube Analytics contains graphs and reports that help you track the performance of your videos and the channel as a whole. There is a huge amount of data contained in various reports (e.g. Watch Time, Traffic Sources, Demographics).

Where is the "YouTube Analytics" section located?

To open YouTube Analytics reports, go to this link or do the following:

View data in YouTube Analytics

Below you will find basic information on how to view and understand Various types data in YouTube Analytics reports.

On the computer

Data filtering

Many reports have filters at the top to make it easier to find the information you need. Data can be filtered by:

• content;
• device type;
• location;
• time range;
• content type ("Video" or "Playlists");
• user status;
• type of playback (broadcast or video on demand), if possible;
• YouTube product (if possible);
• content language.

Line chart Multiple Line Chart

Using a diagram of several lines, you can study the dynamics of views of videos and channels, as well as find out which regions viewers watch your channel more often than others. You can compare up to 25 indicators at the same time.

By default, the top five positions are selected, represented by lines of different colors. However, you can choose to view any content that interests you.

In addition, the selected items can be compared with the total value. To do this, check the box Summary data in the upper right corner.

Note. It is impossible to add other indicators in such a chart.

Combined histogram

An area chart (combo bar chart) shows how selected data relates to total values. This makes it easy to compare performance and identify trends.

By default, the histogram displays the five metrics with the highest values. You can compare up to 25 indicators at the same time.

Pie chart

You can display up to 25 indicators on a pie chart. Hover your cursor over a sector to see what it corresponds to.

Note. In this chart, you cannot look at the time dynamics and compare a single indicator with the final data. To do this, use a multi-line chart or a combined histogram.

bar chart

In a histogram, data is presented in the form of horizontal bars, the length of which is proportional to the value of the indicator. You can't see the time dynamics on this chart. To do this, use a line chart.

Note. This chart also cannot compare an individual metric to a total.

Video analytics- hardware and software or technology that uses computer vision methods for automated data collection based on the analysis of streaming video (video analysis). Video analytics relies on image processing and pattern recognition algorithms to analyze video without direct human intervention. Video analytics is used as part of intelligent video surveillance systems (CCTV, closed circuit television), business management (business intelligence, BI) and video search.

Video analytics functions.

Depending on the goals, video analytics can implement one or several basic functions:

Object detection. As a rule, objects in the camera's field of view are detected using video motion detectors. The main difference between video analytics and IR motion sensors is the possibility of localization (selection) and independent analysis of several objects at once. If motion is not a sufficient feature to localize an object in a frame, then detection can be done using templates. For example, detecting people's faces, car license plates, or detecting slow-moving sea targets.

Object tracking. Tracking (tracking) algorithms make it possible to obtain a partial trajectory of an object’s movement both in the field of view of one camera and a generalized trajectory based on data from several cameras at once. Tracking is necessary to analyze the behavior of an object along its trajectory, for example, to determine the movement of a person against the flow or movement at an increased speed. In addition, tracking is necessary to prevent repeated triggering of video analytics systems on the same objects. Professional systems work according to the rule of “one alarm object - one actuation” to achieve high operator productivity.

Object classification. Some video analytics systems classify objects to filter live notifications or search results. For example, a typical object classifier, using shape characteristics and absolute dimensions, distributes objects into groups: a person, a group of people, a vehicle. More complex classifiers in video analytics systems can determine the gender or return group of a person.

Object identification. Object identification is the most complex component of video analytics systems. Modern systems allow you to identify people by biometric facial features or vehicles by license plates. Identification can be implemented using additional tools beyond video analytics: based on fingerprints, bank card, ticket, pass or mobile device ID.

Detection (recognition) of situations. Video analytics allows you not only to identify objects from streaming video, but also to recognize alarming situations based on behavior analysis of this object, which is not possible to do with a conventional video surveillance system. Also, situational video analytics can automatically detect crossing a signal line, people falling, illegal parking, and a fire.

The results of video analytics are events (messages) that can be transmitted to the video surveillance system operator or recorded in a video archive for subsequent search. In addition, video analytics generates metadata, that is, data structures that describe the content of each frame of a video sequence. Metadata contains information such as the location and identifiers of objects (usually in the form of an alarm frame), the trajectory and speed of movement of objects, data on the separation or merging of objects, data on the occurrence and end of an alarm situation. Metadata is recorded in the video archive and played back along with the video.

Basic types of video analytics.

In terms of application, the following types of video analytics are distinguished:

Perimeter video analytics(perimeter video analytics) is used to protect extended areas and perimeters, detect intrusion and cross signal lines in a “sterile area”. The peculiarity of perimeter video analytics is that violations are relatively rare (hence the zone is called “sterile”), but the shape and type of the object cannot be clearly defined, for example, a person may be crawling or riding a bicycle.

Situational video analytics(situation video analytics) is used to recognize alarming situations related to human behavior or vehicle movement. Situational video analytics can work based on user-defined rules (for example, prohibited parking in a given zone), or based on accumulated statistics (for example, detecting twice as many people in a park as usual at this time of day and on this day of the week).

Biometric video analytics(biometrical video analytics) is used to identify and track individuals based on biometric facial features. Classic biometrics uses black and white lists to compare images of people. Biometric video analytics can work in more complex scenarios, for example, profiling people or comparing observations from multiple cameras in a geographically distributed surveillance network.

Number video analytics(number plate reading) is used to recognize vehicle registration plates, as well as to analyze their movement using data from multiple cameras.

Multi-camera video analytics(multiple camera tracking video analytics) is used to track objects using multiple cameras. The result of multi-camera video analytics is the trajectory of an object on a plan of the entire observation area.

Video analytics system architectures.

From the point of view of hardware and software architecture, the following types of video analytics systems are distinguished:

Server video analytics involves centralized processing of video data on the server. Typically, the server analyzes video streams from multiple cameras or encoders at central processor(CPU) or on GPU(GPU). The main advantage of server video analytics is the ability to combine video analytics algorithms on one hardware platform. The main disadvantage of server-based video analytics is the need for continuous video transmission from the video data source to the server, which creates a load on communication channels.

Built-in video analytics (edge ​​video analytics) implemented directly in the source of video data, that is, in cameras in encoders. Embedded video analytics runs on a dedicated processor (DSP, ASIC, FPGA, ARM or x86 architecture) of the device and transmits results (metadata) along with the video stream. The main advantage of built-in video analytics is to reduce the load on communication channels and on the video data processing server. If there are no objects or events, the video is not transmitted and does not load communication channels, and the processing server does not decode the compressed video for video analysis and indexing. Compared to server-based video analytics, built-in video analytics allows you to increase the efficiency of using communication channels and servers by 10-100 times.

Distributed video analytics is a hybrid solution between server and embedded video analytics, in which processing is distributed between the video data source and the central equipment. For example, in multi-camera tracking systems, object detection is carried out in the video data source, and comparison of results between several sources is carried out on the server.

Ways to identify alarming situations.

Distinguish following methods video analytics settings for automatic recognition of alarm situations:

Rules-based video analytics(rule-based video analytics) is the most common type of video analytics, in which the user pre-defines rules that define alarm events. For example, the user can set a restricted zone where people are not allowed to be or a restricted zone for parking cars.

Video analytics based on statistical learning(statistical learning video analytics) accumulates statistical data on the behavior of objects and generates an alarm in case of non-standard behavior.

General benefits of video analytics.

The main advantage of video analytics over conventional video surveillance systems is the automatic extraction of metadata from the video data stream without operator participation. The resulting metadata can be used to quick search in the video archive, sending alarm notifications and collecting statistics.

Compared to “manual video surveillance,” video analytics can reduce the cost of video monitoring and the human factor in terms of detection and response time. Since a significant part of the video data (more than 99%) in video surveillance systems is not of interest to users, video analytics can radically reduce the load on communication channels and the archiving system by filtering unnecessary video data.

Existing problems.

The main problem with many video analytics implementations is the high rate of false positives, which quickly reduces the economic impact of the technology. The problem is gradually being solved by improving video analysis algorithms, automatic testing on special test benches, and ranking events by importance.

Another problem is the significant cost of system integration and implementation of video analytics. The role of this factor is decreasing due to the emergence of open standards such as ONVIF, simplifying calibration procedures and setting up video analytics.

International standards.

Intelligent video surveillance in general and video analytics in particular are in the early stages of standardization. The following international organizations are developing data transmission and control interfaces for video analytics systems:

ONVIF (Open Network Video Interface Forum) unites about 500 manufacturers and has a strong position in Russia, Europe and Japan. Describes the interfaces for interaction between IP cameras, video analytics servers, video recorders, access control systems (ACS) and other components. Distinctive feature ONVIF is the use of the SOAP protocol for interaction between security system components.

PSIA (Physical Security Interoperability Alliance) unites about 65 manufacturers and has a strong position in North America. A distinctive feature of PSIA is the use of the POST protocol. Compared to ONVIF, the PSIA interface is simpler to implement, but less flexible and scalable.

CAP (Common Alerting Protocol) is a protocol for transmitting messages about alarm situations. Unlike ONVIF and PSIA, the CAP interface is not specialized for video surveillance and video analytics.

Areas of application of video analytics.

The economic effect of introducing video analytics is more noticeable in large, geographically distributed video surveillance networks. As technology becomes cheaper, video analytics is becoming attractive to households and small businesses.

Video analytics is a system of program logic based primarily on video analysis, allowing you to highlight the most important points in the behavior of people or other controlled objects.

Basic principles underlying analytical algorithms:

1. A person cannot cover all the information.

2. A person will never be able to foresee all problematic situations.

3. A person will still make mistakes.

It is necessary to clearly distinguish that SpetsLab video analytics is not based on object detector , as is done by the overwhelming majority of companies, but only uses Object Detector as a component, and not the most significant element, of which there are a lot in the system.

The highest degree of practicality is based on VIDEOSEMANTICS.

Video analytics "Attention!" is built on a self-learning architecture, which, based on previous behavior statistics, determines what is new, unusual, special, unusual in the actions of controlled objects.

As a rule, all criminal phenomena are committed in a non-standard way, because... require special movements, unusual trajectories of movement, specific sequences. They are most often accompanied by psycho-emotional arousal, which forces them to perform unusual actions. All this, as well as any new changes in the frame, is found by the video analyst “Attention!”

"Attention!" is required not only to search for crime, but also allows you to find other important moments in the life of an enterprise, home or street area. Among other things, video analytics makes it possible to very quickly view large volumes of recordings, highlighting only noteworthy moments. Hundreds of hours easily fit into a few minutes of analytical frames.

Video analytics in traffic control

Video analytics also occupies a worthy place in monitoring traffic, identifying accidents, violating traffic rules, protecting and searching for vehicles.

Video analytics “Attention!” - it's just a convenience, not a panacea for all ills. Instead of constantly and every second monitoring the mass of monitors, she suggests only occasionally paying attention to the indicated moments, which, moreover, occur sequentially on one monitor, and only on 1/10 of it. And, in addition, they do not disappear from the screen for a long time, giving you the opportunity to get distracted, go to the toilet or finish watching a beautiful goal in a football match.

This is an effective remedy to combat human factors problems. The last sentence must be taken literally; we are eliminating the human factor, and not creating miracles of artificial intelligence. From the difference between video semantics and classical video analytics: or.

What is video surveillance really?

There is often a substitution of concepts: video surveillance, alarm, security... Various smart people are trying to replace them with video analytics. This word has not even been introduced into the official dictionary yet, it is not yet in other languages ​​of the world, the technology has not yet been officially recognized by anyone, but in our country they already want to replace everything with it technical means security. And the problem is not the quality of the technology, but the meaning in general. Why remove, cancel, replace video surveillance? It needs to be improved, and not converted into smoke detectors, motion sensors, activity sensors and other things that are not typical for video surveillance. In addition, such sensors themselves, which have existed for a long time, cost an order of magnitude less than video cameras with video channels, and they can work in complete darkness.

What exactly is video surveillance? These are the eyes of a person in the distance. And there is no need to throw dust at them. The main thing in the process of video surveillance is the person. If we throw it away, there is no video surveillance. Video surveillance is the same as a person being in the place where the camera is and seeing everything that is happening. He may be asleep, but he sees. Given its selectivity, it does not react to interference such as swaying branches, birds, etc., and is also able to assess at a glance the adequacy of passing people to an object (albeit not always).

So let’s leave these functions for video surveillance, we won’t replace them with anything, we’ll just make sure that a person can comfortably feel that they are simultaneously in hundreds of places of a large number of video cameras. In addition, he could watch TV or even sleep if he really wanted to, and in general experience as little discomfort as possible. We still won’t be able to break human physiology, our task is to adapt to it, to ensure the safety of the object in the conditions of the attention that Mr. Man is ready to pay. Accordingly, there is no point in sending an endless stream of video channels, messages, warnings - no one will perceive them, simply because it is tiring. The task of video analytics is to find a comfortable connection between the video surveillance system and a person.

Video analytics– this is a kind of conditioner for the operator’s brains; it should not be confused with the brains themselves.

We are not fighting criminals; we have already come up with everything to fight them. To see everything around, there is video surveillance. To protect buildings there are security alarm. There are security guards and police to counteract the attack. We fight against the inattentiveness of the operator, the carelessness of the owner, the drowsiness of the security guard,... - we fight the human factor!