Short Bytes: Chip maker Movidius has unveiled “the world’s first embedded neural network accelerator”. Known as the Fathom Neural Compute Stick, this device could be plugged into a Linux device to allow it to perform functions like image recognition, language comprehension, and pattern detection.
To make this a reality, chip maker Movidius has introduced the Fathom Neural Compute Stick. Dubbed as the “the world’s first embedded neural network accelerator”, this USB stick contains a deep learning processing accelerator.
This stick could be connected to an existing Linux device to increase the efficiency of neural networking tasks 20-30 times. Movidius Fathom performs its task at more than 150GFLOPS, consuming less than 1.2W.
“Featuring a full-fledged Myriad 2 VPU, the Fathom Neural Compute Stick not only enables rapid prototyping, but also delivers high levels of neural network compute to existing devices via a USB port,” tells the explainer on Movidius website.
Apart from looking just like a USB device, Movidius Fathom works just as one. It allows pretty much any Linux device to handle advanced neural networks, which is basically AI’s building block.
Once plugged in, Movidius Fathom allows the PC to perform functions like image recognition, language comprehension, and pattern detection.
Embedded inside this tiny stick is TensorFlow, Google’s machine learning software. This will grant devices like drones, cameras, and robots to run computer vision applications. This sounds like a great tool for developers, researchers, and makers who are interested in making their creations more advanced.
According to Movidius CEO, Remi El-Ouazzane — “It’s going to mean that very soon, consumers are going to be introduced to surprisingly smart applications and products.”
Via the website of Movidius, the “qualified customers” can grab Fathom.