Deep Learning with Google and Movidius
Google wants to progress their work with Neural networks to make smart phones more powerful and are going to partner with Movidius to use their chips for faster processing in their future devices.
Movidius, a California based company is leading in embedded machine vision technology providing artificial vision intelligence to the the next generation devices. Google confirmed partnership with Movidius for powerful image recognition technology direct to the smartphone. Right now, certain device capabilities are depend on cloud based apps and through those apps the device will communicate with the server for different tasks like facial recognition, street signs, and finger prints.
Google will place Movidius’ powerful MA2450 chip inside Android devices and by doing this Google will make it easier to match images like faces and signs in real time without uploading pictures. This new technology gives devices the ability to understand images and audio with better accuracy and speed.
The MA2450 is the most powerful processing unit that Movidius has under the Myriad family and the Myriad 2 Vision processing unit is the first vision processor that’s always on. Currently Google’s Project Tango uses the Myriad VPU processor. Project Tango is one of the most fascinating projects that can do real time 3D mapping and vision using tablets. The chip can execute 2 trillion 16-bit operations per second, and haul massive amounts of data using less power than before at 500 milliwatts.
Check out project Tango!
Source: Google ATAP|Youtube
This technology allows for better security using facial recognition and those really cool retinal scans you see in futuristic movies. The application for this technology isn’t limited to the smart device market, but in others as well such as the banking and medical markets.
Check out the vision of Movidius!