Lesson 4

Teach AI to Recognize Images

Mission: Explore how examples and labels help AI make predictions about pictures.

Mission Introduction

Students investigate how a simple image activity can compare visual examples. They focus on labels, privacy, and testing rather than complex model internals.

What You Will Learn

How image labels describe what a model should look for.
Why clear examples are easier to test than blurry examples.
How lighting, angle, and background can affect results.
Why privacy and permission matter when using photos.

Key Words

Image recognition
Label
Example
Prediction
Confidence
Permission

Quick Multiple-Choice Challenge

Teacher Demo

The teacher compares clear and unclear sample images, then discusses why a model might be unsure.

Open CodeHS Starter Code

Open the starter program, sign in if required, fork or copy the program, and return to this page for the tasks.

Open CodeHS Starter Code

Task Checklist

Run the starter image recognition activity. Choose two safe image labels. Review privacy rules before using any image. Add or select clear examples for each label. Test one clear image for each label. Test one tricky image with a different angle or background. Record one reason the model might be unsure. Improve your examples or labels.

Hints

What images should I use?

Use teacher-approved images or classroom objects. Avoid faces, addresses, name tags, or private screens.

What if the model is unsure?

Check the lighting, background, angle, and whether the label is too broad.

Testing Challenge

Test four images: two clear examples and two tricky examples. Explain what changed between them.

Boss Challenge

Your image activity should use safe examples, label them clearly, and explain why a prediction might be correct or uncertain.

Previous: Lesson 3 Next: Lesson 5