Build a Teachable Machine Image Classifier With Your Kid

Introduce your child to the fundamentals of machine learning through a hands-on, interactive computer vision project.

Most children today are passive consumers of artificial intelligence, interacting with algorithms that recommend videos, filter photos, or answer homework questions. To prepare them for a future shaped by technology, we must help them transition from consumers to creators. By building a custom image classifier, your child will learn exactly how computers learn to "see" the world. Using a free web tool and a standard computer webcam, you can guide them through training their very first machine learning model in less than twenty minutes.

The Power of Hands-On AI Education

For children aged 8 to 16, abstract explanations of neural networks and training datasets can quickly become dry and unengaging. Instead, the most effective way to teach these concepts is through direct experimentation. When a child sees that a computer can instantly distinguish between a toy dinosaur and a toy spaceship because of the specific images they captured, the underlying science becomes tangible and exciting.

To achieve this, we will use Google's Teachable Machine, a web-based tool that allows anyone to create machine learning models quickly and easily. There are no paid subscriptions, no software installations, and no specialized hardware requirements. All you need is a computer with a web browser and a webcam.

Step-by-Step Project Guide

Follow these steps with your child to build a custom image classifier that can recognize different family members, distinguish between different hand gestures, or identify different household objects.

Step 1: Set Up the Project

Open your web browser and navigate to Teachable Machine . Click on "Get Started" and select "Image Project" from the options. Choose the "Standard Image Model" to begin. You will see a clean interface with two default classes (Class 1 and Class 2) and a training panel.

Step 2: Gather and Label Your Data

Decide what you want to classify. A fun and simple starting point is distinguishing between two different toys or household items (for example, a coffee mug versus a water bottle).

• Rename "Class 1" to the name of your first object (e.g., "Mug").
• Click the "Webcam" button. Hold the object in front of the camera and click and hold the "Record" button to capture about 100 to 150 images. Encourage your child to tilt the object, move it closer and further away, and change its angle so the computer gets a complete view.
• Rename "Class 2" to the name of your second object (e.g., "Bottle").
• Repeat the recording process for the second object, capturing a similar number of images.

Step 3: Train the Model

Once you have gathered the data, click the "Train Model" button in the middle column. Explain to your child that the computer is now looking at all the pictures, finding patterns, and learning what makes a mug look different from a bottle. Keep the browser tab open and active while the training process completes (it usually takes less than a minute).

Step 4: Test and Play

Once trained, the "Preview" panel on the right will activate. Hold up one of the objects to the webcam and watch the output bars at the bottom. The model will show a real-time percentage confidence of what it thinks it is seeing.

Have your child try to "trick" the model. What happens if you hold the bottle upside down? What happens if you hold up a completely different object? This experimentation is where the real learning happens, as it reveals the limitations of the data we provided.

Parent & Educator Sidebar

Core AI Concept: Supervised Learning

Supervised learning is a type of machine learning where we teach the computer by giving it examples with labels (like showing it pictures of a mug and telling it "this is a mug"). The computer learns the patterns from these labeled examples so it can recognize new, unlabeled objects in the future.

Conversation Starters:

• "Why do you think the computer got confused when we turned the bottle upside down? How could we fix that?" (Answer: We need to train it with pictures of the bottle upside down too).
• "How do you think a self-driving car uses this kind of technology when it is driving down the street?"
• "If we wanted to train a model to recognize different types of dogs, what kind of pictures would we need to collect?"

Action Steps Summary

• Open the tool: Go to the Teachable Machine website on a computer with a webcam.
• Select two objects: Find two distinct items in your house to classify.
• Capture and train: Record 100+ images of each object and click "Train".
• Test and iterate: Try to trick the model and discuss how to improve its accuracy.
• Use the sidebar: Ask the conversation starters to reinforce the core concepts of supervised learning.

Want to share this weekend activity with other parents or educators?