Myriad X Moves Computer Vision and Deep Learning Down to the Bare-Metal

Cameron Coward
Aug 29, 2017 · 3 min read

In recent years we’ve seen a pretty incredible explosion in CV (computer vision) enabled devices, and the software that powers them. It’s not uncommon to see autonomous robots with multiple high definition cameras using sophisticated CV algorithms for guidance, even on low-cost hobbyist robots. But, that capability comes at a pretty big processing expense. Those HD videos contain a lot of pixels, and even with techniques for optimization that make processing the video more efficient, it still takes a lot processing power to do all of the complex analysis we want our robots to do.

The Myriad X VPU delivers AI at the edge for drones, robotics, smart cameras, VR/AR, and more. (📷: Intel)

To try try and jump ahead of these increasing demands, Intel has just announced Movidius Myriad X, which is the first vision processing unit (VPU) that has a dedicated Neural Compute Engine built into the chip. The Neural Compute Engine is a narrow AI (artificial intelligence) that uses deep learning to assist in CV tasks, like detecting specific objects or spatial navigation. Deep learning has recently gained popularity in this field, as it improves results over time, saves time while “training” the system, and takes some of the burden off of programmers to explicitly code every possible scenario.

The Myriad X has a dedicated Neural Compute Engine built into the chip. (📷: Intel)

To try try and jump ahead of these increasing demands, Intel has just announced Movidius Myriad X, which is the first vision processing unit (VPU) that has a dedicated Neural Compute Engine built into the chip. The Neural Compute Engine is a narrow artificial intelligence that uses deep learning to assist in CV tasks, like detecting specific objects or spatial navigation. Deep learning has recently gained popularity in this field, as it improves results over time, saves time while “training” the system, and takes some of the burden off of programmers to explicitly code every possible scenario.

Of course, running a neural network is itself a processor-intensive task. To keep performance high and power low, Intel is making these as low-level as possible. It’s running on the bare-metal, as the expression goes. By keeping these processes below abstraction layers, Myriad X is able to deliver some impressive performance. For instance, it’s capable of taking and processing 8 full-color HD video feeds at a time, and can utilize up to 450GB of internal bandwidth per second.

A look at the VPU’s tiny form factor. (📷: Intel)

This could be a really exciting new development in the computer vision arena, and we’ll be keeping you updated as more details become available.

Hackster Blog

Hackster.io, an Avnet community, is the world’s largest network for hardware & software developers. With 1 million members and 17,000+ projects, beginners and professionals can learn and share how to build robotics, industrial automation systems, AI-powered machines, and more.

)
Cameron Coward

Written by

Author, writer, maker, and a former mechanical designer. www.cameroncoward.com @cameron_coward

Hackster Blog

Hackster.io, an Avnet community, is the world’s largest network for hardware & software developers. With 1 million members and 17,000+ projects, beginners and professionals can learn and share how to build robotics, industrial automation systems, AI-powered machines, and more.

Welcome to a place where words matter. On Medium, smart voices and original ideas take center stage - with no ads in sight. Watch
Follow all the topics you care about, and we’ll deliver the best stories for you to your homepage and inbox. Explore
Get unlimited access to the best stories on Medium — and support writers while you’re at it. Just $5/month. Upgrade