Student Help Sheet
This page shows you how to use Teachable Machine step by step. Use it whenever you get stuck!
Download the printable version
You can also download this as a Word document:
Download Help Sheet (DOCX)How to Use Teachable Machine
1. Open Teachable Machine
- Open Google Chrome or Microsoft Edge.
- Type teachablemachine.withgoogle.com in the address bar and press Enter.
- Click the "Get Started" button.
- Click "Image Project".
- Choose "Standard image model".
2. Set Up Your Classes
- You will see Class 1 and Class 2 on the left.
- Click the pencil icon next to "Class 1" to rename it (e.g. "Circle").
- Rename "Class 2" as well (e.g. "Triangle").
- To add more classes, click "+ Add a class" at the bottom.
3. Upload Your Images
- Under each class, click "Upload".
- Click "Choose images from your files".
- Navigate to your folder and select all the images for that class.
- You can select multiple images at once by holding Ctrl and clicking each one, or press Ctrl+A to select all.
- Repeat for each class.
4. Train Your Model
- Once all classes have images, click the big "Train Model" button in the middle.
- IMPORTANT: Do NOT switch tabs or minimise the browser while training!
- Wait for the training to finish. You will see a progress bar.
- When it's done, the Preview panel will appear on the right.
5. Test Your Model
- In the Preview panel on the right, you can choose how to test.
- Click "File" tab, then "Choose images from your files".
- Select a test image. The model will show its prediction and confidence percentage.
- Record the result in your results table.
- Test with several images to see how accurate your model is.
Tips for Drawing Good Training Images
Do:
- Use a white background for all images (unless told otherwise)
- Draw with black lines so the shape is clear
- Make each image slightly different — change size, position, thickness
- Draw at least 10 images per class
- Save as PNG format
Don't:
- Don't draw all images exactly the same — the model needs variety
- Don't use different coloured backgrounds for different classes (unless testing bias)
- Don't make images too tiny — the shape should fill most of the canvas
- Don't save as JPEG if possible — PNG keeps the image clearer
Frequently Asked Questions
My model won't train — what do I do?
Make sure you have uploaded images to at least two classes and each class has at least 2 images (aim for 10+). Check you haven't accidentally left a class empty. Try refreshing the page and uploading again.
The training gets stuck or freezes.
Don't switch to another tab while training — this can pause it. If it freezes completely, refresh the page (F5) and start again. Your images won't be saved on the site, so you'll need to re-upload them.
My model gets everything wrong!
Check these things:
- Did you put the right images in the right class? (e.g. circles in Circle, not Triangle)
- Are your images clear enough? Small or blurry images confuse the model.
- Do you have enough images? More is better — try 15–20 per class.
How do I save my model?
Unfortunately, Teachable Machine doesn't save automatically between sessions. If you close the browser, you'll need to re-upload and re-train. However, your images are saved on your computer, so you can quickly re-upload them.
Can I use the webcam?
In this unit, we are not using the webcam. We use uploaded images instead. This is so we can control exactly what the model sees and learn about datasets properly.
How do I add a third class?
Scroll down below Class 2 and click the "+ Add a class" button. You can add as many classes as you like!
What does the percentage/confidence mean?
The percentage shows how confident the model is in its prediction. 95% means the model is very sure. 55% means it's not very sure and could easily be wrong. Higher is better!