The ultimate all-seeing surveillance system could soon be here with thanks to Nvidia. The company’s new venture is called Metropolis and is an integration like no other between CCTV and complex machine learning artificial intelligence. It’s designed to be able to continuously process and contextualize data collected from over a billion cameras and it could be hitting our streets as early as 2020.
Metropolis has the ability to analyze and interpret 30 billion images per second. One of the great things about the system is it could be used to create autonomous drones that can stream video back from the sky, or be used as a security robot looking for any trouble that may be lurking. This system could also be used as the next wave of security camera recording panoramic views of our streets. No longer would police need to rely on eye witnesses for evidence, as cameras would track the entire scene as it happened.
Some of the other benefits of the system include using deep learning techniques to help redirect traffic and track what customers are buying in busy supermarkets. There would be no need to store data for later use as it would constantly be analyzed by Nvidia’s machine learning systems which will include vehicle recognition, face recognition, and both road and pedestrian traffic patterning.
A company once famous for its gaming PC’s and graphics cards, Nvidia is now using artificial intelligence to take it to the next machine learning level. Some of its previous AI successes include creating a ‘watch and learn’ system for self-driving cars to understand how traffic works, and the development of software for industrial security drones.
BriefCam is one of Nvidia’s partners, and last year a video was released that showed how its security software tracked vehicles and individuals. Things such as shoplifting or someone jumping a security fence will be picked up and passed to a human operator for further analysis. The footage is then combined to create a shortened review video that highlights all the important events. This could save banks and supermarkets no end of time when needing to review CCTV footage.
