Build a Teachable Machine Image Classifier With Your Kid

Issue #32 • March 13, 2024

Most children experience artificial intelligence as a passive consumer, whether they are watching recommended videos or asking a chatbot to write a story. To truly prepare the next generation for an AI-driven world, we must shift their perspective from consumers to creators. By building a custom image classifier, kids aged eight to sixteen can demystify how computers learn to see. Using a free, web-based tool and a standard webcam, you can help your child train their very first machine learning model in less than thirty minutes.

Demystifying Computer Vision

When a self-driving car stops for a pedestrian or a smartphone unlocks via facial recognition, they are using computer vision. To a child, this can feel like magic. In reality, it is simply a mathematical model trained on thousands of images. To teach this concept, we will use Google's Teachable Machine, a free, highly visual platform that requires no coding, no paid subscriptions, and no specialized hardware.

In this project, you and your child will build a Rock-Paper-Scissors classifier. The goal is to train the computer to recognize these three distinct hand gestures through the webcam and accurately predict which gesture is being shown in real time. This hands-on exercise provides a clear, tangible demonstration of how machine learning models are built, trained, and tested.

Step-by-Step Project Guide

Follow these steps with your child to build, train, and test your custom image classifier:

Step 1: Set Up Your Classes

Open a web browser on any computer with a webcam and navigate to Teachable Machine . Click on "Get Started" and select "Image Project", then choose "Standard Image Model". You will see a interface with two default classes: Class 1 and Class 2. Rename Class 1 to "Rock" and Class 2 to "Paper". Click the "Add a class" button to create a third class and name it "Scissors".

Step 2: Gather Your Training Data

Click the "Webcam" button under the "Rock" class. Have your child hold their hand in a fist (the "Rock" gesture) in front of the webcam. Click and hold the "Hold to Record" button. Capture about one hundred to one hundred and fifty images. While recording, instruct your child to slightly rotate their hand, move it closer and further from the camera, and shift it to different areas of the frame. This variety is crucial because it helps the model learn the shape of the fist rather than just a single static position. Repeat this exact process for the "Paper" class (flat hand) and the "Scissors" class (two fingers extended).

Step 3: Train Your Model

Once all three classes have at least one hundred images, click the "Train Model" button in the middle column. It is vital to keep the browser tab open and active during this process. The computer is now analyzing the pixels in the images, finding patterns that distinguish a fist from a flat hand or extended fingers. This process typically takes less than a minute.

Step 4: Test and Debug

Once training is complete, the "Preview" panel on the right will activate, showing a live feed from your webcam. Have your child make a "Rock" gesture. Look at the output bars below the video. Does the model confidently predict "Rock" with a high percentage? Try "Paper" and "Scissors".

If the model gets confused, this is a perfect learning opportunity. If it misidentifies a gesture, look at the training data. Did your child hold their hand too close to the camera? Was the background lighting different? Have your child record fifty more images for the confused class, retrain the model, and test it again.

Parent & Educator Sidebar

Core AI Concept: Supervised Learning

Supervised learning is the process of training an AI model using labeled data. In this project, the labels are "Rock", "Paper", and "Scissors". The computer does not inherently know what a hand is, it simply looks for pixel patterns associated with each label we provided.

Conversation Starters for Your Kid:

• "What happens if we try to show the camera a completely different object, like a coffee mug? Which class does the computer guess, and why do you think it makes that guess?"
• "If we only trained the model using my hand, do you think it would work just as well for your hand? Why is it important to have diverse training data?"
• "How do you think a self-driving car is trained to recognize a stop sign versus a speed limit sign using this same technology?"

Action Steps Summary

• Access the Tool: Open Teachable Machine on a computer with a working webcam.
• Create Three Classes: Set up classes labeled "Rock", "Paper", and "Scissors".
• Record Training Data: Capture at least one hundred webcam images for each class, ensuring varied hand angles and distances.
• Train and Test: Click "Train Model" and use the live preview panel to test the accuracy of your model.
• Discuss the Concepts: Use the sidebar questions to help your child connect this simple game to real-world AI applications.

Want to explore more practical, family-friendly AI projects and executive workflows?