Build a Teachable Machine Image Classifier With Your Kid

Introduce your child to the core concepts of machine learning by building a custom, real-time visual recognition model.

In an era dominated by discussions about artificial intelligence, teaching children how these systems actually work is one of the most valuable skills a parent can provide.

Instead of just consuming technology, children aged eight to sixteen can easily build their own functional machine learning models in less than twenty minutes. Using a computer with a standard webcam and a free web-based tool, you and your child can demystify AI by training a computer to recognize different objects, hand gestures, or facial expressions. This hands-on project builds critical thinking, logical reasoning, and a foundational understanding of data science.

Getting Started with Teachable Machine

To begin this project, you do not need any coding experience, paid subscriptions, or specialized hardware. You only need a laptop or desktop computer equipped with a working webcam and an internet connection.

Open a web browser and navigate directly to Google's Teachable Machine . Click on the "Get Started" button, and select "Image Project" from the available options. Next, choose the "Standard Image Model" option. This will open the main workspace where you and your child will build, train, and test your custom visual classifier.

Step 1: Define Your Classes and Gather Data

A machine learning model learns by comparing different categories of information, which are called "classes." For this project, we recommend building a simple "Rock, Paper, Scissors" game classifier, or a tool that distinguishes between a favorite toy and a school book.

On the screen, you will see two default boxes labeled "Class 1" and "Class 2." Click the pencil icon to rename them. For example, rename Class 1 to "Rock" and Class 2 to "Paper." You can click "Add a class" to create a third box named "Scissors."

Now, it is time to gather data. Click the "Webcam" button inside the first class box. Have your child hold their hand in a "Rock" fist in front of the camera. Click and hold the "Hold to Record" button. Capture at least one hundred images while your child slightly rotates their hand, moves it closer to the camera, and moves it further away. This variety helps the computer learn the general shape of a fist rather than just one specific angle. Repeat this exact process for the "Paper" and "Scissors" classes.

Step 2: Train the Model

Once you have gathered data for all classes, click the blue "Train Model" button in the middle column. This process takes about ten to thirty seconds.

During this step, explain to your child that the computer is analyzing the pixels in the images to find patterns. It is looking for edges, curves, and colors that distinguish a fist from a flat hand or two extended fingers. Make sure to keep the browser tab open and active while the training completes, as the processing happens directly inside your web browser.

Step 3: Test and Iterate

Once training is complete, the "Preview" panel on the right side of the screen will activate, showing a live feed from your webcam. Have your child make a "Rock" gesture in front of the camera. Below the video feed, you will see progress bars indicating the percentage of confidence the model has for each class.

This is where the real learning happens. Challenge your child to "break" the model. What happens if they use their other hand? What happens if they stand far back in the room? What happens if you, the parent, try the gesture?

If the model gets confused, explain that the computer only knows the specific data it was given. To fix the errors, simply go back to the class boxes, record more diverse images (such as using the other hand or changing the lighting), and click "Train Model" again. This iterative process perfectly demonstrates how machine learning systems improve through better data.

Parent & Educator Guide

Core AI Concept: Supervised Learning

This project demonstrates supervised learning, where an AI model is trained on labeled data (the images you categorized). The computer does not actually "know" what a hand is; it simply matches pixel patterns from the live webcam feed to the patterns stored in its training database.

Conversation Starters:

• "Why do you think the computer got confused when we changed the lighting or used a different hand?"
• "How do you think self-driving cars use this exact same technology to recognize stop signs and pedestrians?"
• "If we only trained the model with my hand, why might it struggle to recognize your hand?" (This introduces the concept of algorithmic bias).

Action Steps Summary

• Access: Go to Teachable Machine and start a new Standard Image Project.
• Label: Create and name your classes (e.g., Rock, Paper, Scissors).
• Record: Use your webcam to capture at least one hundred diverse images for each class.
• Train: Click "Train Model" and wait for the browser to process the patterns.
• Test: Use the live preview to test the model and add more data to fix any mistakes.

Looking for more engaging, family-friendly technology projects?