- General
- February 17, 2019
- 6 minutes read
Xnor Debuts A Miniaturised AI Platform
An Xnor prototype camera, hosting its newly debuted AI solution that operates independently of the cloud or a power source.…
 |
| An Xnor prototype camera, hosting its newly debuted AI solution that operates independently of the cloud or a power source.
image : Xnor
|
Seattle based startup Xnor has unveiled a standalone battery-free, solar-powered AI platform that it says will enable AI to run on in an “always-on mode” on a range of edge devices, like the tiny camera in the above picture. Xnor says the AI solution is able to detect visual objects, including people, making use of deep learning models previously dependent on powerful GPUs but now able to run on a simple solar cell.
The company says the miniaturised platform will enable AI technologies – embedded into solutions ranging from consumer wearables to self-driving vehicles – to scale faster and “more efficiently, and with far less environmental impact”. Its making use of the prototype camera displayed in the above and below photo to demonstrate its capability.
 |
| image : Xnor |
The camera is capable of analyzing video content via deep learning algorithms, Xnor says. It’s able to identify, classify and code objetcs within a frame and then send this information to a recipient in near real-time via wireless communication protocols designed for IoT (internet of things) applications. Photo and video data never leaves the device, Xnor says, enabling security to help guarantee a user’s privacy is protected.
Being small and lightweight, the AI platform would make for easy use in a range of devices, taking up lesser space that can be utilised for other functions. “With technology this low power, a device running on only a coin cell battery could be always on, detecting things every second, running for 32 years!” Xnor Head of Hardware Engineering, Sam Naderiparizi, said in a press statement.
 |
| Xnor co-founder and former CEO Ali Farhadi
image : Xnor
|
According to a study by the U.S. Natural Resources Defense Council, data centers in the U.S. process an ever increasing number of traditional AI algorithms that are estimated to consume close to 140 billion kilowatt-hours of electricity each year by 2020, the equivalent of 50 large coal-fired power plants.
Xnor is hoping to not only have a hand in saving businesses from the estimated $13 billion annual energy cost, but to also curb the release of nearly 100 million metric tons of carbon pollution each year. A demo video of the new AI solution debuted by Xnor is displayed below :
Founded in 2017, Xnor is a relatively young startup whose platform is built on research conducted at the University of Washington and the Allen Institute for Artificial Intelligence, a research hub founded by late Microsoft co-founder Paul Allen. The company is led by Jon Gelsey, the former CEO of web security company Auth0.
It’s backed by nearly $15 million in funding according to Crunchbase data, from investors including Madrona Venture Group, Autotech Ventures and NGP Capital.
