Teachable Machine

Features

• Image Classification – Train a model to recognize different image categories.
• Sound Classification – Train models to recognize short audio samples.
• Pose Classification – Teach the model to distinguish different body positions.
• Webcam Capture – Collect image or pose examples directly from a webcam.
• Microphone Input – Record sounds for audio classification.
• File Uploads – Use existing examples as training data.

Use Cases

• Teaching students basic machine-learning concepts.
• Creating simple image-recognition systems.
• Building gesture-controlled projects.
• Creating voice or sound-controlled applications.
• Developing interactive games.
• Building webcam-based experiments.
• Creating object-classification prototypes.

Dos and Don'ts

Do

• Give each class clear and meaningful examples.
• Include different angles, backgrounds, and variations where appropriate.
• Use enough examples for each class.
• Test the model with examples it did not see during training.
• Add more examples when the model repeatedly makes mistakes.
• Keep training classes reasonably balanced.
• Review how the model performs before using it in a real application.
• Use the tool as a practical way to teach AI and machine-learning concepts.

Don't

• Assume the model understands objects or concepts like a human.
• Train it using only a few nearly identical examples.
• Use heavily biased training examples and expect reliable results in different conditions.
• Treat a high confidence score as proof that the classification is correct.
• Use a simple prototype as a production-ready machine-learning system