How to Scale Performance with Intel® Movidius™ Neural Compute Sticks | AI News

  • 概览
  • 文字稿

Get a brief glance at how hosts Ian Bragg, and Neal Smith discuss how to dynamically scale applications for performance with the Intel® Movidius™ Neural Compute Stick, or Intel® Movidius™ NCS.

See the Neural Compute App Zoo on Github

Watch Dynamically Scaling Performance with Multiple Intel® Movidius™ Neural Compute Sticks

Subscribe to the Intel Software YouTube Channel

Keep up to date with the AI News YouTube Playlist

I'm David Shaw, and this is AI News. This time, we bring you a real-world use case from a recent Intel video series called Deep Learning Made Easy. Hosts Ian Bragg and Neal Smith discuss how to dynamically scale applications for performance with the Intel Movidius Neural Compute Stick, or Intel Movidius NCS. Let's take a look.

In the video, Neal highlights the role of the Neural Compute App Zoo, an online repository of pre-trained neural networks and applications shared by Intel and the community. He demonstrates how adding multiple Intel Movidius Neural Compute Sticks improves the performance of an application that analyzes objects from a street camera. Next, he walks through the details of how the program works. Let's look at Ian Bragg and Neal Smith in action, running the Neural Compute App Zoo and what happens when they add a couple more Neural Compute Sticks.

So let's just go ahead and run it.

I can see that it shows cabs and cars and that kind of thing. But it's running pretty slowly.

So let's go ahead and plug in a few of these and see what we get.

Can you show us what the change will be?

Yeah.

So yeah, these are actually going by at near-real-time speeds.

Pretty impressive, right? Check out the video for yourself in the links. And I will see you next week.