Lesson 1: Your First Model

45–60 minutes Windows PC, Browser, Paint

Today you will…

Equipment Needed

Key Vocabulary

Machine LearningA type of AI where the computer learns from examples instead of being given rules.
ModelThe thing the computer builds after learning — it can make predictions on new data.
Class / LabelA category or group. In this lesson your two classes are "Circle" and "Triangle".
Training DataThe example images you give the computer so it can learn.
PredictionThe model's guess about what it sees in a new image.

Starter Activity (5 minutes)

Think about it: How do you know the difference between a circle and a triangle? Write down 2 differences below.

Discussion points: Link to how computers need to "learn" these differences from examples rather than being told rules. Ask: "Could you write exact rules that always work?" — this leads to why ML is useful.

Main Tasks

Part A: Create Your Training Images

  1. Create a folder on your desktop called TM_Lesson1. Inside it, create two folders: Circles and Triangles.
  2. Open Microsoft Paint from the Start menu.
  3. Draw a circle using the shape tool. Use a white background and black outline. Make the circle fill most of the canvas.
  4. Save the image as circle_01.png in your Circles folder. Use File > Save As > PNG.
  5. Draw at least 10 different circles. Make each one slightly different — change the size, thickness, or position. Save each one with a different number (circle_02, circle_03, etc.).
  6. Now draw at least 10 triangles in the same way. Save them as triangle_01.png, triangle_02.png, etc. in your Triangles folder.

Tip: Circulate and check students are varying their drawings. Common issue: students draw identical shapes — remind them variation helps the model learn. Aim for 10–15 per class minimum. Faster students can draw 20+.

Part B: Build Your Model on Teachable Machine

  1. Open your browser and go to teachablemachine.withgoogle.com. Click "Get Started".
  2. Click "Image Project", then choose "Standard image model".
  3. You will see two classes: Class 1 and Class 2. Rename Class 1 to Circle and Class 2 to Triangle.
  4. Under Circle, click "Upload" and then "Choose images from your files". Navigate to your Circles folder and select all your circle images.
  5. Do the same for Triangle — upload all your triangle images.
  6. Click the big "Train Model" button. Wait for training to finish (do not switch tabs!).
  7. Once trained, the Preview panel will appear on the right. Click "Choose images from your files" in the preview and test with one of your images. Does the model get it right?

Part C: Record Your Results

Test your model with 5 images (mix of circles and triangles). Record the results below:

Test Image Actual Shape Model's Prediction Confidence (%) Correct? (Yes/No)
1
2
3
4
5

Common Mistakes & Quick Fixes

Challenge / Stretch Task

Draw 5 "tricky" images — shapes that are hard to classify (e.g. a very rough circle, a rounded triangle, or a shape that's half-way between). Test them with your model. What happens? Write your findings below.

Exit Ticket

Expected answers:

  1. Machine learning is when a computer learns from examples/data instead of being given rules.
  2. Circle and Triangle.
  3. So the model can learn what different circles/triangles look like — more variety = better at recognising new ones.

Lesson Downloads

Demo Datasets

Pre-made circle & triangle images

Download ZIP

Help Sheet

Step-by-step Teachable Machine guide

Download DOCX
← Home Lesson 2 →