A terrifying vision of future cities: Video reveals how AI will team up with CCTV to analyse your every move

  • Nvidia's Metrolpolis camera platform combines CCTV with artificial intelligence
  • The system can constantly process data from over a billion CCTV cameras 
  • It could be used to streamline traffic and track supermarket shoppers
  • But Metropolis could also be used to create autonomous security robots 

By HARRY PETTIT FOR MAILONLINE

Published: | Updated:

Nvidia is striving to create a creepy, all-knowing surveillance system unlike any other.

The company's 'Metropolis' intelligent video analytics system combines always-on CCTV with complex machine learning artificial intelligence.

The platform can constantly process and contextualise data from a network of over a billion cameras, the company claims.

In a new video, Nvidia outlines how its enormous camera system could be used in cities as early as 2020. 

Scroll down for video 

Nvidia's Metropolis is an intelligent video analytics platform that can process footage from over a billion security cameras looking for patterns and usable data. This artist's impression reveals how the Metropolis system tracks multiple people and objects at once

Nvidia's Metropolis is an intelligent video analytics platform that can process footage from over a billion security cameras looking for patterns and usable data. This artist's impression reveals how the Metropolis system tracks multiple people and objects at once

WHY WOULD CITIES WANT TO USE NVIDIA?

Nvidia's 'Metropolis' technology can link up over a billion cameras across a single city.

Instead of security footage needing to be stored it would be analysed by Nvidia's machine learning systems.

This would include facial recognition, vehicle recognition, and road and pedestrian traffic patterning.

For normal city camera systems, hiring personnel to watch countless video feeds is impractical.

But Nvidia claims that its system, which takes advantage of artificial intelligence and machine learning, will cut security labour costs.

Advertisement

Using data from cameras dotted across a city, warehouse or supermarket, Metropolis could analyse and interpret 30 billion images a second, totalling 100 trillion images an hour.

The system could be used to create autonomous drones that stream video back from the sky, as well as security robots that seek out trouble, NewAtlas reports.

Metropolis could also help create the next generation of security cameras: Ultra-high definition systems that record panoramic views of a street.

This would allow the cameras to track an entire scene as it unfolds instead of having to focus on individuals.

In the new video, Nvidia outlines that its system could use deep learning to help redirect traffic at busy intersections and parking lots.

It could also track what customers are buying in crowded supermarkets.

'Nvidia's intelligent video analytics platform changes how video is analysed and turns it into insight,' the company said in the video.

'It can be used for real-time analysis in applications like multi-object classification, facial recognition, and behaviour analysis.

'This one-AI platform could improve travel, find a lost pet or child, drive business insights and success, and keep us safe from harm.'

Nvidia's Metropolis platform encompasses several of the company's products. The firm says the system could be used to streamline traffic and multi-story parking garages in busy cities

Nvidia's Metropolis platform encompasses several of the company's products. The firm says the system could be used to streamline traffic and multi-story parking garages in busy cities

Security footage captured by Metropolis would not need to be stored for later analysis like traditional systems

It would instead be constantly analysed by Nvidia's machine learning systems.

This would include facial recognition, vehicle recognition, and road and pedestrian traffic patterning. 

Nvidia, based in Santa Clara, California, once made graphics cards for gaming PCs.

But its artificial intelligence software is now taking the firm toward machine learning systems.

The company has already helped to create AI systems in self-driving cars that 'watch and learn' to understand how traffic works.

It has also developed software for industrial drones that tracks assets on large-scale work sites.

One of Nvidia's partners, BriefCam, released a video last year that shows its security software tracking individuals and vehicles through security footage.

It will flag up unusual behaviours, such as someone shoplifting or jumping a security fence, which can then be passed to a human operator for analysis.

The system then combines this footage to create a quick review video which shows all noteworthy events in a given time frame happening at once.

This condensed format could save supermarkets and banks time when reviewing CCTV footage.

But the system can narrow down security footage further than this - it can be told to only show the red cars that passed along a stretch of road over the course of a day, or to only show bicycles that turned left at a certain intersection.

AI

Most watched News videos