Picture Detective
See how a computer works out what's in a picture.
What you'll learn in this level
- That a computer only sees pictures as numbers
- How pixels turn into recognised things
- What the confidence number really means
How Computers Understand Pictures
A computer doesn't see shapes the way we do. First, every picture is broken into a huge grid of tiny coloured dots — pixels — stored as numbers.
From those numbers the AI looks for patterns, recognises objects like chairs, trees or dogs — and tells you how sure it is about each guess.
How it works
- 1
Pixels
The picture is split into tiny coloured dots. They are called pixels. Each pixel is just a number for the computer.
- 2
Patterns
The AI looks for edges, shapes and colours that belong together.
- 3
Objects
From the patterns it finds whole things. A chair, for example, or a dog.
- 4
Confidence
For every find the AI says how sure it is. That number is called confidence.
Look at a few scenes. You will see what the AI finds. You will also see how sure it is.
Guess first, then experiment
The image AI says: "Cat, 98% sure". Does that mean it is really almost certain?