The Calliope Teachable Machine is a web-based application that lets you easily create your own machine learning model. To do this, images are collected with the camera and assigned to different classes. Based on this training data, a model is created that can recognize and classify inputs from the camera. calliope
Calliope Teachable Machine offers the option to connect the Calliope mini directly. This allows the models you create to be used in simple programs to develop, for example, interactive projects or prototypes. calliope
Data quality
How does a machine learn to distinguish things?
How do training data influence the result of an AI?
What problems can arise when using AI?
How it works
Prepare
The Calliope mini is connected to the computer via Web-USB and camera settings are defined. makecode.calliope
⚠️ Calliope Teachable Machine requires a browser that supports Web-USB, e.g. Chrome or Edge. makecode.calliope
Generate data
In the first step, name a class and save the name. In the next step, add images to this class.
Hold down the button to capture images (10 images/second).
The more images a class has, the better that class can be recognized. There should be at least 60 images per class!
A training model requires at least three classes. These two steps must be repeated for at least two additional classes.
Train the model
Based on the collected data, a model is created that can distinguish between the classes.
Model details
The model details provide information about accuracy and reliability.
Programming
The model can be processed directly in MakeCode and thus integrated into prototypes with the Calliope mini.
Transfer the program to your Calliope mini via Web-USB so that in the next step you can connect your Calliope mini.
Apply and review the model
The algorithm’s decisions are displayed directly and can therefore be checked and reflected on.
Handling generated data
The data generated while working with Calliope Teachable Machine are stored locally in the browser.
When a project is saved, all associated data – including the recorded images – are packaged into a .zip file and stored locally on the device. calliope
You can find great Teachable Machine examples at LOFI Robot:
Teachable Calliope App by LOFI Robot
