Skip to content
BENIBIX Learn

Image Classification

Also known as Image Recognition, Image Labeling

AI gives a whole photo one label, such as cat or dog.

Think of it like

It is like sorting photos into boxes. One box for cats, one for dogs, one for birds.

Example

A farm uses image classification on photos of apples on a belt. Each apple gets the label good or bruised, and the bruised ones are pushed off the line.

Why it matters

It is the base of most vision work. If you only need to know what a picture shows, this is the cheapest way.

Definition

Image classification is when an AI model looks at a whole picture and gives it one label, such as cat or dog. The model gives each label a score for how sure it is, and the top score wins. It says what the picture shows, but not where the thing is. It is the simplest vision task, and the first one deep learning got very good at. It sorts photos, checks parts in factories, and flags skin spots for a doctor.
The whole photo gets labels with scores. The top one wins.

Common mistake

Using it when you need to know where things are, or how many. One label for the whole picture cannot tell you that.

Key takeaways

  1. 1
    One label. For the whole picture.
  2. 2
    With a score. How sure it is.
  3. 3
    What, not where. No position.

Compare

Learn it properly