Kipod Server IP-videoserver

2020: Integration of heatcameras into the intellectual Kipod platform

The Synesis company, the resident Klastera of information technologies of Skolkovo Foundation integrated heat chambers into the intellectual platform of "the smart city" Kipod. Now a system can be used for control of spread of a koronavirusny infection. On June 3, 2020 Skoloko's Fund reported about it.

Using thermal imagers the platform in real time analyzes temperature of people in a flow and announces responsible services deviations from the set parameters. The sensing technology of persons, in turn, defines to whom the diseased contacted. The platform also estimates density of accumulation of people, signals about non-compliance with a social distance and collects statistics about attendance of the place – for example, about the peak periods and the maximum load.

Thus, use of artificial intelligence will allow to solve three fundamental problems at once:

• prevent visit of public places by sick people and carriers of a virus (including without strong indications of a disease);
• it is effective to separate flows of healthy and sick people;
• to quickly define the group of people, contacting to the diseased.

If business centers, infrastructure facilities (the airports, stations, the subway, etc.) and trade malls have an opportunity in large quantities to check temperature of all entering people, then distribution of one of the most dangerous infections of the 21st century can slow down considerably. We excluded a human factor in work with thermal imagers – people with the increased temperature are defined by cameras, and the artificial intelligence helps to systematize the obtained information and to react quickly to it. It is effective and does not require big labor costs. Integration with the system of face recognition – one more huge step towards monitoring automation. And we also made it,

For June 3, 2020 starts a pilot project for large network of drugstores in Florida (USA). Using the Kipod platform the personnel will be able to separate flows of buyers for those who have a normal body temperature, and those who come with the increased temperature and also to control regulations of social distancing.

When the whole world is in pandemic conditions, such solutions as at Synesis company are the most relevant and have the top priority. Really, if people are sure that at visit of public places - whether it be drugstore or football stadium - their relatives will be most protected from contacts with already sick, then peace will be able to be restored from pandemic effects much quicker. I am very glad that solutions of our residents are found interest not only in Russia, but also abroad,

2018

According to information for March, 2018 the platform of intellectual video surveillance and monitoring of public security of KIPOD from Synes company allows to increase efficiency of law-enforcement and municipal services by 70%.

KIPOD integrates an unlimited number of video cameras from any objects. The platform processes data in real time and notifies the staff of law-enforcement and municipal services on non-staff situations.

"For example, to trace the criminal, it is enough to police officer to load an identikit into KIPOD. The platform will analyze data from video cameras and will issue all similar people less than in 5 seconds. At the same time the technology allows not only to find suspects according to the photo from the database of law enforcement authorities, but also to make trajectories of their movement on the city". Bondarenko Sergey, CEO of Synes LLC

KIPOD will recognize not only persons of people, but also numbers of machines in a moving flow. Using the platform law enforcement officials obtain also information on the stolen cars, violations of traffic regulations, traffic jams and accidents.

The technology of the Skolkovo company is already used by law enforcement authorities of Belarus, Russia, Kazakhstan and Azerbaijan for process automation of monitoring of public security and prevention of crimes. KIPOD allows to reduce costs of the state on disclosure of crimes to 30%.

2014

Kipod and technology of ranging of events

On January 24, 2014 the Synes company announced results of tests of technology of ranging of events in Kipod video surveillance system.

The technology of ranging of events gives the chance to the operator of a video surveillance system - to concentrate on the analysis of more significant events by means of automatic purpose of priorities and ranging of data, it significantly increases effectiveness of work. The efficiency increases:

• uses of disk space on local and central servers,
• search in archive as storage time of video data is defined by degree of their importance.
• requirements to the speed of data transmission channels decrease that allows use of mobile clients with slow connection.

The technology of ranging was tested at office of Synes company and in the adjacent territory. Data arrived from 16 cameras at the same time within one hour. Three operators independently of each other using three different configurations were engaged in data processing:

• without video analytics (only the detector of the movement);
• using video analytics;
• using video analytics and ranging of events.

As video analytics the detector of intersection of signal lines, the detector of the movement and the detector of the left subject in a signal zone and also classification of objects by type were used.

As the equipment were used:

• Server of video analytics KP-SERVER-100
• Workstation of video analytics KP-DESKTOP-100
• Камеры Ganz PixelPro ZN-MD221M, Axis P3346, Dallmeier DF4910HD-DN

As a result of tests mean values of volumes of the transferred and written data, time of activity and reaction of the operator, the number of the missed events are calculated. The maximum quantity of the arrived events was 1644 (for the detector of the movement), the volume of the transferred data was 101647 Mb, the volume of the written data – 98352 Mb.

Results of testing:

Extrapolation of data retrieveds on the situational center shows advantages of use of technology of ranging of events: if 10 objects, on each of objects on 10 cameras are the share of one operator, and each camera generates 10 events per hour, then for an hour the operator needs to process 1000 events from 100 cameras.

It is obvious what without use of video analytics of people cannot monitor such number of cameras at the same time. And even when using video analytics, to the operator one notification in 4 seconds will come, on average. But, as showed researches, about 90% of the arriving events are not really important. Using the mechanism of ranging of data on importance it is possible to lower load of the operator practically by 10 times.

In this case ranging of events allows to improve a situation cardinally: the operator will receive the notification only on important events (about 100 per hour), direct participation of the operator will be required only approximately time in 40 seconds that enough time for adequate assessment of the situation and adoption of the correct solution leaves. At the same time requirements to concentration of attention of the operator decrease. It leads to reduction of a response time of the operator and also practically to exclude cases of the admission of important events.

The priority of an event can depend on different signs: from type of an event, a zone of observation, accuracy of recognition, range of an object and others and also from a combination of these factors.

The report shows performance improvement of work of the operator at 40-50 times in comparison with the systems of "manual" video surveillance without video analytics and by 5-6 times in comparison with the systems of intellectual video surveillance with analytics without use of technology of ranging.

In this testing of technology of ranging provided advantages:

The gain of performance of operators can be used, both for increase in the performed work, and for a staff decrease of employees when preserving the current volumes of tasks.

The module from VisionLabs is switched on

On February 3, 2014 the Synes company announced end, in cooperation with VisionLabs, works on the module of recognition of license plate numbers.

The system of intellectual video surveillance Kipod developed by Synes is created by joint stock company the open platform of the videonalatiki and allows integration of software modules of other suppliers of software solutions.

Cooperation with VisionLabs company allowed Synes to add quickly the module of recognition of the state license plate numbers and to expand functionality of a video surveillance system of Kipod for the systems "Safe City". The choice of an algorithm VisionLabs is caused by high high-speed performance on video of the HD format and compatibility with Linux.

The module promotes recognition of automobile license plate numbers of Russia, Kazakhstan, Belarus, Jordan and the European Union countries.

Among module advantages:

• the built-in analysis algorithm and corrections of results guarantees steadily high probability of reliable recognition of license plate numbers in actual practice;
• the algorithm of parsing of results provides the minimum quantity of false operations;
• intellectual adaptive algorithms cause lack of need for recognition settings;
• algorithms of character recognition and search of area with license plate number provide low computing resource intensity;
• video analytical modules work in separate process and are automatically restarted in case of hangup.

Existence of a videoarchive helps with quick search of car numbers in the database that gives the chance quickly to identify the violator. Integration of a system with billing of a parking reduces influence of a human factor, increasing efficiency of personnel and making impossible abuse of office.

2013

Integration into the cameras Basler

Cameras of machine vision of Basler company are integrated in the summer of 2013 in IP видеосервер by Kipod Server of Synes company. Modern coders, such as H.264 AVC, reduce the volume of the transferred video data in hundreds of times. Under favorable conditions for video surveillance and a slow-moving scene of observation, compressed video is indistinguishable from uncompressed for a human eye, and losses of useful information are not considerable.

At adverse conditions of video surveillance (insufficient illumination, meteorological rainfall, broad dynamic range) or strongly changeable scene, the coder can enter significant distortions (artifacts) to the transferred video signal. Algorithms * identifications of objects are the most sensitive to these distortions, including identification algorithms on:

• to biometric signs of the person;
• to license plate numbers of the car.

The systems of video analytics processing uncompressed video allow to eliminate influence of the video coder and to reach higher level of accuracy of recognition.

Thanks to joint work of Synes and Basler companies, IP видеосервер Kipod Server is integrated with cameras of machine vision via the GigE Vision interface allowing to transfer uncompressed video on Gigabit Ethernet network.

Except video analysis IP видеосервер implements a high-performance compression (coding) of uncompressed video on the graphic processor (GPU). Compressed video or results of work of video analytics can be written in archive and are transferred with IP видеосервера to attendant workstation or to other video surveillance system (VMS) via the ONVIF interface. Thus, IP видеосервер can be the instrument of integration of biometric or number cameras into a normal video surveillance system.

Cameras of machine vision of Basler company are optimized for video analytics and industrial application. It should be noted the following advantages of cameras of machine vision in comparison with normal IP cameras:

• lack of artifacts of compression of H.264 and MJPEG;
• use of more professional components (high-quality management system, optics, sensor);
• transfer of an uncompressed video flow of the high definition (HD);
• smaller geometrical and color aberrations in a frame for measurement of parameters of subjects to observation;
• high frequency of the personnel for tracking objects in a dense flow.

Except GigE Vision, IP видеосервер supports other digital interfaces for acceptance of uncompressed video - USB and HD-SDI.

The joint solution of Synes and Basler is aimed at the following tasks:

• selection of persons from a passenger traffic for the subsequent identification of the personality on biometric signs;
• recognition of license plate numbers of cars and cars;
• detection of objects at protection of perimeter at a big distance in difficult conditions;
• high-quality video fixing of production or technology process, including for the analysis of parameters of a product, sending, banknotes;
• reading of one-dimensional and two-dimensional barcodes.

Thus, the joint solution of Synes and Basler promotes increase in efficiency and expansion of opportunities of video analytics of high definition due to use of cameras of machine vision with good optical characteristics and high definition of the image including in difficult conditions of video surveillance.

Search of the people connected with the left or carried away objects

Given versions of the detector of the left/carried away objects and the client application Kipod Desktop implement function of quick search of the people connected with the left or carried away objects.

In case of a disturbing situation "the left subject" or "the carried-away subject", the operator of a system of video analytics can see freeze frames at once and watch video with people who were near a subject. The new version of the detector of the left objects steadily works in places of mass accumulation of people and more precisely predicts time of emergence of the left subject. This time mark is used for context search of people in the field of a subject. Thus, the operator in the shortest possible time (10 seconds) can quickly find "owner" of the left subject or "thief" of the carried-away subject.

The detector of the left/carried away objects passed tests in pilots zones of Moscow and St. Petersburg Metro. Scope of the detector are "The safe city", transport infrastructure, strategic and production facilities.

Detector of a fight and non-standard behavior of people

The detector of a fight (fight) developed by Synes company can be used as a part of the systems of city video surveillance within the project the Safe city. The detector reacts to sharp or periodic movements and is intended for integration into the systems of security video surveillance on the basis of the international ONVIF interface. Works in structure IP видеосервера of Kipod Server of Synes company and it can be applied in a complex with other detectors: accumulations of people, quality control of video, the left objects, fire and smoke.

At the heart of an algorithm of detecting of fights and antisocial behavior the following fact lies: a fight is characterized by existence of the sharp disordered movements which are performed without considerable movement on a scene. The algorithm can be separated into several stages:

• At the first stage detection of the fast changings happening on the stage is performed. At the same time the techniques allowing to minimize influence of variability of color and dimensional responses of an object are applied.
• At the second stage there is an accumulation in time of the signal received at the first stage. As a result those sharp movements which occur without considerable movement of objects (locally amplify. Also the fast movements connected with movement of objects, for example, run are cut.

The signals received as a result of the second stage, are compared to an adaptive threshold. Its level depends on the general activity on the stage. As a result, as showed testing, detection of fights and other non-staff situations happens rather reliably.

For detection of antisocial behavior, along with video analytics, the audioanalytics configured on detecting of shout and sharp intonations can be used.

Thus, the detector of a fight allows:

• to automatically monitor all arising emergency situations connected with antisocial behavior, fights, hooliganism and fights;
• increase efficiency of response of security services to similar emergency situations and also prevent their undesirable development.