This first episode of season three defines reference implementations, shows you how to use them, and identifies what hardware to choose when developing your own IoT solutions.
Hello. My name is Martin Kronberg. Welcome to the IoT Developer Show, season three. In this season, we take a closer look at reference implementations developed by Intel for developers.
In this episode, I will let you know what reference implementations are, how they work, what hardware you can use to deploy them, and how we can use them in developing your own IoT solutions.
So you might be wondering—what are reference implementations? Well, Intel's reference implementations [sic] are open source, end-to-end solutions, which address specific use cases. Things like detecting if workers are in a restricted zone using computer vision, or an intelligent vending machine that can monitor sales, temperatures, and maintenance logs.
The reference implementations are built and validated using Intel developer tools.[sic] They're a great way to demo how an IoT solution works using Intel technology and can be a great resource in building your own IoTsolutions. If you want to run these implementations, you have a wide range of hardware options.
What hardware you choose will depend on the compute requirements of the reference workload and the performance that you need. For some of the lighter weight workloads, you can use the UP Squared* AI Vision Developer Kit, which runs on an Intel Atom® processor.
For the implementations that use multiple neural network models for computer vision, you want to use something more powerful like the IEI TANK* AIoT Developer Kit with an Intel® Core™ i7 or an Intel® Xeon® processor, especially if you want to see a smooth frame rate from the output.
Now, what is the best way to use these implementations? You can start by looking at all the implementations that are currently available on the Intel® Developer Zone and learn a bit more about the one that you want to use. From there, go to the GitHub* page in order to learn more about the specifics of how the implementation works and instructions on how to get it running on your hardware.
Finally, we encourage you to clone or fork the repository in order to leverage the open source project and make it your own. And that's the basic overview. In the next two episodes, we take a close look at a couple of implementations.
A store traffic monitor using a retail environment, and an object flaw detector used inside industrial environments. We show some demos, dig into the code, and help you get started with these two implementations. As usual, all the links are provided, and I'll see you guys next time.