Build a Teachable Machine Image Classifier With Your Kid

Children today are surrounded by artificial intelligence, from the recommendation algorithms on video platforms to the voice assistants in our homes. However, most children only experience AI as passive consumers, viewing it as an opaque, almost magical technology. To prepare them for a future deeply integrated with these tools, we must demystify how they work. The best way to do this is by building a real machine learning model together, using nothing more than a standard computer webcam and a free web browser tool.

This project uses Google's Teachable Machine, a web-based platform designed to make machine learning accessible, visual, and highly interactive. There are no paid subscriptions, no coding requirements, and no specialized hardware setups. In this activity, you and your child will train a custom image classifier that can instantly recognize different hand gestures, toys, or household objects.

The Project: Build a Hand Gesture Detector

We will build a model that can distinguish between three distinct states: a thumbs-up gesture, a thumbs-down gesture, and a neutral state where no hand is visible. This simple setup provides an immediate, visual demonstration of how computers learn to see.

Step 1: Set Up the Workspace

Open a web browser on your computer and navigate directly to https://teachablemachine.withgoogle.com/ . Click on the "Get Started" button, then select "Image Project" followed by "Standard Image Model". This opens the training interface, which is divided into three clear columns: Gather Data, Train Model, and Preview.

Step 2: Create and Name Your Classes

In machine learning, a "class" is a category of things the computer is trying to recognize. By default, the interface shows two classes. Have your child rename "Class 1" to "Thumbs Up" and "Class 2" to "Thumbs Down". Click the "Add a class" button at the bottom of the column to create a third category, and name it "Neutral".

Step 3: Gather the Training Data

This is where the hands-on fun begins. Click the "Webcam" button inside the "Thumbs Up" class box. Have your child sit in front of the camera, hold up a clear thumbs-up gesture, and click and hold the "Record" button.

Instruct them to move their hand slightly, tilt it, and change their distance from the camera while recording. This variety is crucial because it teaches the computer to recognize the gesture from different angles. Collect about 100 to 150 image frames. Repeat this exact process for the "Thumbs Down" class, and finally for the "Neutral" class, where your child should simply sit in front of the camera without showing any hand gestures.

Step 4: Train the Model

With the data gathered, click the "Train Model" button in the middle column. It is important to leave the browser tab open and active during this process. The training takes about ten to fifteen seconds. Explain to your child that the computer is currently looking for patterns, such as the shape of the hand, the position of the fingers, and the contrast against the background, to build a mathematical rulebook for each class.

Step 5: Test and Debug

Once training is complete, the "Preview" column on the right will activate, showing a live feed from the webcam. Have your child make a thumbs-up gesture. They will see the percentage bar for "Thumbs Up" instantly jump to 100%.

Now, try to break the model. What happens if you use your left hand instead of your right hand? What happens if you stand far back in the room? If the model gets confused, explain that this is not a failure, it is a data gap. You can easily go back to the first column, record more diverse images, and retrain the model to make it smarter.

Parent & Educator Sidebar

Core AI Concept: Supervised Learning

Supervised learning is the process of training an AI model by showing it labeled examples. The computer does not inherently know what a hand is. It simply analyzes pixels, finds commonalities among the images in each category, and uses those patterns to make predictions on new, unseen images.

Conversation Starters:

• "If we only trained the model using my hand, why might it struggle to recognize your hand?" (Introduces the concept of algorithmic bias and data diversity).
• "How do you think self-driving cars use this exact same technology when they are driving down the street?" (Connects the project to real-world computer vision applications like stop signs and pedestrians).
• "What other categories could we train this model to recognize next?" (Encourages creative problem-solving, such as sorting recycling or identifying pet behaviors).

Action Steps Summary

• Go to https://teachablemachine.withgoogle.com/ on a computer with a webcam.
• Create three classes: "Thumbs Up", "Thumbs Down", and "Neutral".
• Record 100 to 150 webcam frames for each class, ensuring varied angles and distances.
• Click "Train Model" and wait for the browser to complete the training process.
• Test the model in the preview window and discuss how the computer identifies the patterns.

Looking for more engaging, family-friendly AI projects to build together?