Teachable Machine with Scratch Tutorial: Connect Recognition Outputs with Program Events
Teachable Machine with Scratch connects a recognition model's outputs with actions in a block-based program. A compatible bridge or extension supplies those outputs, while Scratch logic decides how scores trigger events. Class mappings and repeated-input behavior matter because a noisy prediction stream can produce unintended rapid changes in the interactive project.
Here's a popular Teachable Machine with Scratch video — watch it, then summarize it with one click. Or paste any other YouTube link below.
— or paste your own YouTube link —
Want to keep the key points from a YouTube video? Paste the link below to turn it into structured notes.
Why Use Teachable Machine with Scratch Tutorial: Connect Recognition Outputs with Program Events
- Identify the bridge that supplies predictions to Scratch
- Map model classes to program behavior
- Distinguish scores from trigger decisions
- Handle repeated or fluctuating outputs
- Test the complete recognition-to-action interaction
Frequently Asked Questions
Does Scratch automatically load every Teachable Machine export?
Integration requires a compatible bridge or extension with its own interface. Identify that implementation and model format instead of assuming any exported artifact can be used directly by ordinary Scratch blocks.
Why map class labels before connecting actions?
The model's classes define output meaning, while program actions define the interaction. An explicit mapping prevents a changed class order or name from triggering behavior intended for another prediction.
What can a threshold contribute to the interaction?
A threshold defines when a score is strong enough to trigger a program event. Its behavior should be tested with representative inputs rather than treated as a guarantee that the recognized class is correct.
Why manage fluctuating predictions?
Successive frames can produce changing scores even for a nearly unchanged input. Program logic needs appropriate timing or stability handling so those fluctuations do not create unintended rapid repeated actions.
How should the project be tested?
Test inputs through model recognition, bridge output and Scratch behavior together. This reveals integration and timing problems that are absent when the classifier or block program is examined alone.