2.1.8 Running a Model On Your Embedded Device

This guide describes how to run your model on your embedded device. To do this, generating a Knowledge Pack is required.

Knowledge Packs

A Knowledge Pack takes the event detection model you generated in your pipeline and transforms it into a file that can be run on your hardware device at the edge. When the Knowledge Pack is on your device, it starts outputting classification IDs that correspond to your events of interest. You can download a Knowledge Pack for your platform of choice on the Download Model page. You can find out more about integrating Knowledge Pack APIs into your firmware by referring to the section Knowledge Pack Firmware Integration.

  1. Open the Download Model page.

  2. Select your platform. This tutorial uses SAMD21.

  3. Useful information about the model, like Estimated Memory Usage or Latency before download, displays.

  4. Select Library for “Format” and Custom for “Data Source”, then click DOWNLOAD.

  5. Integrate the Knowledge Pack APIs into your embedded device firmware by following the Knowledge Pack Integration.

Connecting To Your Device

When a Knowledge Pack is flashed onto your device, the classification results can be viewed by connecting to the virtual COM port in a terminal application. Follow the steps below for viewing the recognition results:

  1. Open a terminal emulator such Data Visualizer Terminal.

  2. Set your serial speed to the baud rate of your device.

  3. Create a new connection to the serial port.

    You will now see classification results printed in the terminal.