NVidia started selling object recognition software to automakers
The manufacturer of graphic processors entered the market of automation for the automotive industry. Nvidia began selling hardware and computer vision software to Bentley, Aston Martin, Tesla and Rolls-Royce. Tools will help onboard computers to recognize objects.
The computer vision system for unmanned vehicles from Nvidia consists of a graphics processor , which is installed in a vehicle with software for object recognition, and a server hosted in the automaker's laboratory. On serial cars, the system is not installed: automakers work with prototypes on which 12 cameras are installed. Training system will take two years, write Vedomosti .
The graphics processor processes the images from 12 cameras installed in the car and presents them as groups of pixels, the software searches for group boundaries, compares with reference images, and estimates the probability of identity. Danny Shapirro, director of automotive technology, Nvidia, is confident in the ability to create a system that will recognize objects in the same way as humans. The system will have to recognize unknown images: "This is important, because if, say, a machine that does not have information about moose, sees an elk on the road, it still needs to make a decision to slow down and dodge the collision with an animal."
')
Automakers and Google are now working on the bases for “in-depth training”, which is preparing to launch a mobile car on the roads this summer.
Head nVidia Ren-Sun Huang and head of Tesla Ilon Musk
The computer vision system for unmanned vehicles from Nvidia consists of a graphics processor , which is installed in a vehicle with software for object recognition, and a server hosted in the automaker's laboratory. On serial cars, the system is not installed: automakers work with prototypes on which 12 cameras are installed. Training system will take two years, write Vedomosti .
The graphics processor processes the images from 12 cameras installed in the car and presents them as groups of pixels, the software searches for group boundaries, compares with reference images, and estimates the probability of identity. Danny Shapirro, director of automotive technology, Nvidia, is confident in the ability to create a system that will recognize objects in the same way as humans. The system will have to recognize unknown images: "This is important, because if, say, a machine that does not have information about moose, sees an elk on the road, it still needs to make a decision to slow down and dodge the collision with an animal."
')
Automakers and Google are now working on the bases for “in-depth training”, which is preparing to launch a mobile car on the roads this summer.
Head nVidia Ren-Sun Huang and head of Tesla Ilon Musk
Source: https://habr.com/ru/post/290836/
All Articles