ABC 2026

This section proposes an exercise to provide hands-on experience with a tool like EdgeImpulse for designing TinyML applications. As an end device, we will use your smartphone. You also need an Edge Impulse account; it is time to create it!

Image classification

based on: https://docs.edgeimpulse.com/tutorials/end-to-end/image-classification

In this tutorial, we’ll build a model to distinguish between two objects. We will use a coin and a pen, but feel free to pick two other objects of your choice!

Setting up the project

You first have to create a project by clicking on Create new project:

Check that, on the bottom right, “one label per data item” is selected:

Building a dataset

To make your machine learning model “classify images,” you must capture a lot of images of these objects as examples. When training the model, these example images let the model distinguish between them. In this tutorial, we’ll build a model that can differentiate between two objects: a coin and a pen.

Capture the following amount of data - make sure you capture a wide variety of angles and zoom levels:

• 20 images of a coin
• 20 images of a pen
• 20 images of the background where you are taking the images

So go to the section Data acquisition then Collect data and click on “Connect a device”:

and “Scan QR code to connect to your phone”

On your phone click on “Collecting images” and start taking and labeling each picture according to the object: