Nvidia Metropolis

Nvidia Metropolis is the analytical platform for work with video provided in 2017.

Nvidia Metropolis gives opportunities for processing and the analysis of video data and is intended for creation of the Smart City systems on the basis of artificial intelligence technologies. The platform is based on products and Nvidia technologies and serves for creation of solutions in spheres of public security, management of traffic flows, fight against crime and many others.

2017

Nvidia began to advance the Russian IT products for creation of the smart cities

In September, 2017 Nvidia reported that the Russian companies "Vocord" and VisionLabs – developers and producers of intelligent systems video surveillances and face recognition, became her partners within the program of creation of solutions on the Metropolis platform. It included, in particular, the program of Vocord of biometric face recognition of high accuracy and the cross-platform solution of face recognition VisionLabs.

Around the world for participation in this Nvidia program selected only 12 companies. It officially recommends their applications for use in projects of creation of the safe and smart cities.

Within partnership of the company will get priority access to innovations of Nvidia and will get support in promotion of products in foreign markets: Nvidia is going to include information on these solutions in the maketingovy materials at different mzhdunarodny exhibitions and conferences, told TAdviser in the company.

Vocord and VisionLabs appeared among the first partners of Nvidia in the program of creation of solutions on the Metropolis platform

Inclusion in the program means that solutions of partners were tested with hardware platforms of Nvidia and correspond its requirement in all respects, explained TAdvser in the company.

Performance of creation of the biometric Vocord FaceControl templates - the most resource-intensive transaction - on GPU NVIDIA Tesla P40 does not exceed 3.1 ms on one image. The company says that such characteristics allow to create large distributed systems with the most low power consumption and cost in terms of one camera. And use of cameras of recognition of VOCORD NetCam with the built-in system of creation of templates based on the Nvidia Jetson platform gives the chance to cardinally reduce costs for transfer of images and reduces load of networks and servers.

In turn, the VisionLabs technologies using the hardware-software Metropolis tools are applied in different spheres, including bank – to control of a time recording of employees and access to work files on the person, fight against fraud, confirmation of the identity of clients and transactions in remote links of service, and process automation; retail – for data collection for marketing statistics, fight against theft and increase in the service level of clients; and many other directions of business.

All partners of the program:

• Briefcam is the Solution for the fast analysis and search in video, notifications in real time and collecting of quantitative data.
• Deepvision AI - Identification of brands, logos and products for accomplishment of analytics and providing the personalized content to retail clients.
• Herta Security - Solutions of face recognition in places of accumulation of people for law enforcement bodies.
• Icetana is the Self-training analytical solution for detection of anomalies in real time from surveillance cameras and the instant notification of operators.
• Ironyun is Search in video using AI algorithms using a natural language interface.
• Omni AI - A self-training system for search of anomalies in video, images, in a cyberspace, SCADA and data analysis in real time.
• OpenALPR is the Program of automatic license plate recognition for the smart cities, law enforcement bodies and the companies.
• Sensen Networks - Solutions for a parking and control of speed for the cities. Solutions for casino, for example, for the analysis of board games and clients.
• Sensetime is Solutions of face recognition for public places, shops and the systems of access.
• VisionLabs is the Cross-platform solution of face recognition for controlled (biometric authentication and reliability of the identification information) and uncontrollable scenarios (shops, the smart cities and public places).
• Vocord is the Program of biometric face recognition of high accuracy.
• Xjera is Solutions for calculation of people and transport for the commercial organizations, differing in a fine precision, the high level of customization and protection.

Platform presentation

On May 8, 2017 the Nvidia company submitted the analytical platform for work with video Nvidia Metropolis.

The technology helps to make the cities safer and more smart by means of application of algorithms of deep learning to video flows in applications for ensuring public security, management of traffic and resource optimization.

The serious analysis of video which in real time turns dry data into valuable information allows to provide deep learning, increasing security and improving quality of life. The NVIDIA Metropolis platform allows to apply AI to each video flow to do the cities it is more smart. Deepu Talla, the vice president and the CEO for solutions of Tegra in NVIDIA

As a part of the Metropolis platform several Nvidia technologies.

The high-speed output on end devices is executed on the basis of the computing platform for the built-in NVIDIA Jetson systems, on servers and data centers the analysis happens on NVIDIA Tesla accelerators. Data visualization is executed by means of professional graphics of NVIDIA Quadro. All complex platform is supported by packets of tools for developers from NVIDIA, including JetPack, DeepStream and TensorRT.

Complex analysis of video in real time, and these data — the largest source of information allows to provide deep learning. Hundreds of millions cameras are installed in commercial buildings, government and public institutions, public transport, on streets of the cities already now, and by 2020 it is expected that the total quantity of cameras in the world will reach one billion.

Only small part of the written video data is analyzed, and the most part is just stored for future analysis if it is required (someone will request records on specific number and time). Attempts of the analysis of video data in real time were too difficult and unreliable until recently, in comparison with viewing videos by people. Though people can pass some parts which specially trained AI can notice. Application of algorithms of deep learning in cameras, video recorders, on servers and in clouds allows to solve a problem of permanent monitoring of video with high accuracy.

Support of an initiative of Nvidia of creation of the cities with AI constantly grows, already more than 50 companies in the field of creation of the smart cities, including: Hikvision, Avigilon, Dahua, Milestone and Hanwha Techwin, suggest the clients to obtain necessary information from video flows by means of algorithms of deep learning, and this processing is based on hardware-software solutions of Nvidia company.

Potential at the Metropolis platform rather big, it can be applied to video flows to creation of intellectual applications in different scopes, beginning from transport and finishing with commercial tasks. Advantage of deep learning using the powerful graphic processors Nvidia is that data can be analyzed much quicker and more precisely. It is important to provide the necessary speed of response to the events on streets of the cities.

Robotics

• Smart cities (Smart cities, Smart cities)