The name of the base system (platform): | Vocord FaceControl |
Developers: | Vocord |
Last Release Date: | 2018 |
Technology: | Cybersecurity - Biometric identification, Video surveillance systems, Systems of video analytics |
Content |
2018: Vocord NetCam.AI
In March, 2018 the Russian company Vokord, the developer of intelligent systems of video surveillance, biometric identification and computer vision – put on the market VOCORD NetCam.AI face recognition camera. Nvidia Jetson TX1 constructed on the platform of artificial intelligence, VOCORD NetCam.AI will become the first-ever camera which independently, without participation of the server, will recognize persons. All process, including detecting of persons, creation of biometric templates and comparison happens completely "onboard". Use of this technology several times reduces costs for the server hardware, load of data networks and energy consumption of a system of recognition in general.
VOCORD NetCam.AI is the solution for creation of large distributed systems of security based on face recognition and the intellectual video analysis in scales of the city or region and also on subjects to mass stay of people: at stations, in the subway, at the airports, on sports arenas and so forth. In the commercial sector the camera can be used to the analysis of consumer behavior, protection against illegal acts, control of production processes and also as a component of Internet of Things.
In NetCam.AI the modern model of Edge-calculations is implemented: video data are processed directly where they appear, i.e. in the camera. The camera completely undertakes resource-intensive image processing, removing the most part of computing loading from servers. The camera takes up to 25 persons in one frame that allows to use it on the most highly loaded objects.
Implementation of face recognition in the camera allows to lower load of network and the server hardware. Calculations show that when using NetCam.AI of costs for servers decrease on average by 10 times. At the same time the general energy consumption of a system based on NetCam.AI decrease approximately twice. And because the camera transfers only a biometric template which weighs less than 1 KB to the server, also amount of data, transferred on network is sharply reduced. The requirement to network transmission capacity becomes about 100 times less, than for the traditional systems where the video flow entirely is broadcast. Small amount of data allows to transmit them even through mobile network 3G/4G and WiFi.
The camera is steady against interruptions in network functioning: in case of communication loss, biometric templates remain in the buffer in the camera; after connection recovery they will be sent to the server. Data processing in the new camera happens on the GPU module of the NVIDIA Jetson TX1 platform. In NetCam.AI own neural network algorithm of face recognition VOCORD FaceControl which is among the most exact based on the international NIST and MegaFace tests is used. Besides, the camera supports loading of different neural network algorithms and can be used both for biometric identification on the person, and for recognition and the analysis of a broad spectrum of events and objects: car numbers, types of vehicles, education of crowd, behavior of people, etc. The new camera, as well as other devices of a line VOCORD NetCam, is optimized for obtaining the image of the best quality. So, the matrix of high resolution (5 MT), broad dynamic range, fine tuning of parameters of shooting under a zone of interest (ROI), 12-bit coding of a signal and the built-in algorithms of improvement of quality allow to identify reliably persons and other objects at a great distance and with the broad range of illumination.
The camera is expected use both in premises, and on the street. It works with the broad range of temperatures (from - 25C ° to + 50C °) and has a class of protection against climatic factors of IP67. The first public demonstration of the camera VOCORD NetCam.AI will take place at the end of March at a conference on the accelerated calculations and solutions for artificial intelligence of NVIDIA GTC 2018. And in May acceptance of preorders on new devices will begin.
2017: Vocord NetCam FC
The Vokord company announces a concept of the innovation camera of face recognition of Vocord NetCam FC in September, 2017. This camera is development which contains the built-in system of detecting of persons and calculation of biometric templates. This camera is capable to distinguish persons without contact with the server and to check them on locally loaded (or stored on the server) base of images. Implementation of face recognition and tracking directly in the camera allows to lower load of network and the server hardware and to build large-scale distributed systems of face recognition on such large objects as the subway, stations, the airports, stadiums, shopping centers, etc. and also in solutions of the class "Safe City".
In the camera of Vocord NetCam FC own neural network algorithm of recognition of Vocord which based on testing of a NIST and MegaFace is the world leader in work accuracy is applied. High-speed performance of this algorithm is provided by the built-in Nvidia Jetson platform thanks to which the camera is capable to build more than hundred biometric templates per second. Such high-speed performance is necessary for the systems of mass noncooperative face recognition.
This camera, as well as other devices of a line NetCam, has several advantages to which the image quality improvements which are built in an algorithm belong, automatic optimization of parameters of the camera depending on an interest zone, low requirements to network transmission capacity if search is run not locally, and on the server (the volume of a biometric template ≈ 0.5 Kb).
2016: The cameras Vocord NetCam will be completed with the lenses Tamron
In June, 2016 Vocord and Tamron company signed the agreement on cooperation. All line of specialized cameras for video surveillance and video analytics of VOCORD NetCam for application in the field of capture/control of persons of people and license plate numbers of vehicles will be completed now, including, and the lenses Tamron.
Let's note that the cameras VOCORD NetCam complete hardware-software complexes which function on large commercial, sporting and industrial facilities and also in large-scale projects of the class "Safe City". The lenses TAMRON allow to receive the minimum overall dimensions when preserving the quality inherent to large-size lenses. Thanks to it also possibilities of application of video surveillance systems and analytics will extend that it is especially important in interiors where to apply big lenses happens it is impossible.
The cameras VOCORD NetCam will be presented in projects of any class that will allow to increase considerably a market share of products of Vokord company in a domestic segment of video surveillance and video analytics and also to enter the new markets, first of all the European.
2012
The Vokord company updated a line of video cameras of new generation of VOCORD NetCam4 developed especially for application in the systems of biometric video surveillance at the beginning of 2012. The functionality for work in the conditions of uncontrollable lighting is added to video cameras, new opportunities of automated management of external devices and are improved the parameters affecting image quality.
Requirements to video cameras for pattern recognition systems are significantly higher, than possibilities of normal CCTV cameras. And cameras of "machine vision" with very high resolution unreasonably roads for such tasks. The face recognition of the VOCORD NetCam4 video camera developed especially for systems meets all necessary requirements and at the same time is more available at the price, than foreign analogs to the systems of "machine vision". VOCORD NetCam4 has no domestic analogs.
In VOCORD NetCam4 low-noise precision ATsP with digit capacity to 14 bits are used (WDR up to 84 dB). Functions of automatic adjustment of exposure with precisely set minimum and maximum time of exposure, conversion function of digit capacity of the image from 12 bits to 8 bits with automatic adaptation to change of conditions of illumination and preserving of effective dynamic range, an automatic white balance, exact formation of control signals of the video sensor are implemented. VOCORD NetCam4 in the automatic mode manages the motorized lenses that allows to configure a diaphragm and focusing with a high accuracy (P-Iris technology). Depending on tasks of the video camera are completed with different types of video sensors: both black-and-white, and color, with the resolutions from 1 to 11 Mpiksel and technology of the central lock "global shutter". All this allows to achieve images of very high quality. For increase in accuracy of recognition of the video camera issue 2 independent video elementary streams: uncompressed (RAW) and compressed (MPEG).
In a new line of video cameras work with external devices is improved: the interface of exchange with controllers is added by the ACS and other actuators, management of rotating devices, window washers and window wipers, pulse lighters is implemented. Function of synchronous processing of audioinformation is added to video cameras: a stereo pair of linear audio inputs with a connectivity of miniature electret microphones with phantom power supply and a stereo pair of audio outputs to which loudspeakers are directly connected. Thus, VOCORD NetCam4 video cameras are the universal front-end of the device managing all periphery of a boundary of control in the systems of biometric video surveillance.
All advantages of new models of VOCORD NetCam4 video cameras in full are implemented in the solution of Vokord company - the system of face recognition VOCORD FaceControl.
The VOCORD FaceControl system is intended for security in the subway, at stations and other transport hubs, on points of customs and border control and also in public places: on sporting and mass public events. A system works in the noncooperative mode – i.e. selects and will recognize persons of the people moving in a flow. It allows to apply successfully it in places of massy accumulation of people where it is impossible to force them "pose" in front of the video camera.
The scalability of the VOCORD FaceControl system allows to create the full-function geographically distributed systems of biometric identification. Such system from several territorial and spaced boundaries of control since 2012 works at railway stations of Ryazan.