Have you ever wondered how an AI system can look at a picture and identify what it sees?

 

In this activity, we will use Teachable Machine to create a simple AI model that can recognize whether an image contains a Car or a Bike.

 

No advanced coding is required!

​

What is Teachable Machine?

​

Teachable Machine is a beginner-friendly tool that allows students to train a machine-learning model using examples.

Instead of writing complicated code, we can show the computer different examples and allow it to learn patterns from them.

For our project, we will teach the AI to recognize two categories:

🚗 Car
🏍️ Bike 

The AI will learn from the images we provide and then try to identify new images it has never seen before.

 

Our AI Project: Car vs Bike

The Challenge

Imagine you are creating a smart traffic system.

A camera captures vehicles on a road. We want our AI system to look at the image and predict:

Is it a Car or a Bike ?

To create this project, we need to teach our AI using examples.

 

Step 1: Create the Classes

First, create two categories for our project.

Class 1 – Car 🚗

Collect several different pictures of cars.

Try to include:

• Small cars

• SUVs

• Sedans

• Cars from different angles

• Cars in different backgrounds

 

Class 2 – Bike🏍️

Collect different pictures of two-wheelers.

Include:

• Motorcycles

• Scooters

• Different models

• Different viewing angles

• Different backgrounds

The more useful and varied examples we provide, the better our model may learn the visual differences.

 

Step 2: Collect Training Data

Now we give examples to the AI.

For the Car class, add several car images.

For the Bike class, add several bike images.

​

Step 3: Train the AI

Once enough examples have been collected, click the Train Model option.

Teachable Machine processes the examples and creates a machine-learning model.

The model tries to learn visual patterns that help distinguish one class from another.

For example, it may learn patterns related to:

• Shape

• Structure

• Wheels

• Overall appearance

• Visual features

​

Step 4: Test the Model

Now comes the exciting part!

Give the trained model a new picture that was not used during training.

For example, show it a picture of a car.

The model might predict:

🚗 Car – High confidence

Then show it a picture of a bike.

It might predict:

🏍️ Bike – High confidence

Try different images and see how the model performs.

 

Step 5: Can AI Make Mistakes?

Absolutely!

Suppose we show the model a difficult image:

• A bike partly hidden behind a car

• A vehicle photographed from an unusual angle

• A very blurry image

• A crowded road

• An image with poor lighting

The AI may make an incorrect prediction.

This teaches an important lesson:

AI is not perfect.

The quality and variety of the training data can affect how well a model performs.

 

Step 6: Improve Your Model

If your AI makes mistakes, don't worry!

Ask:

Did we provide enough examples?

Were the images varied enough?

Were some images unclear?

Did the training data represent different situations?

Add better examples and train the model again.

This is an important part of learning AI:

Test → Find Problems → Improve → Test Again

 

Where Can This Idea Be Used?

Our simple Car vs Bike project demonstrates an idea called image classification.

Similar AI techniques can be used in many real-world applications.

For example:

🏥 Identifying certain medical images
🌱 Recognizing healthy and unhealthy plants
♻️ Sorting different types of waste
🐾 Identifying animals
📦 Recognizing products
🚦 Understanding objects in traffic environments

Real-world systems are much more complex, but the basic idea of learning from examples is similar.

 

What Students Learn

Through this activity, students learn:

✅ What Artificial Intelligence means
✅ What Machine Learning means
✅ How AI learns from examples
✅ What training data is
✅ What a machine-learning model is
✅ What image classification means
✅ Why data quality matters
✅ Why AI can make mistakes
✅ How models can be improved

​

🧩 YoungAI Challenge

Can You Make the AI Better?

After creating the Car vs Bike model, try changing the project.

Can you create an AI model that recognizes:

🚗 Car vs 🚌 Bus

🏍️ Bike vs 🚲 Bicycle

🐶 Dog vs 🐱 Cat

🍎 Apple vs 🍌 Banana

🌱 Healthy Plant vs 🍂 Unhealthy Plant

Start with two classes and experiment with different examples.

​

💡 Think Like an AI Engineer

Ask yourself:

What problem am I trying to solve?

What data do I need?

What examples should I give the AI?

How will I test it?

What can I do if it makes mistakes?

These questions are part of the process of developing useful AI systems.

​

                                  Happy Learning