Decoder. plain-English AI glossary

Image Classification

● Core

Assigning an image to a category or class - is this a cat, dog, bird, or car?

Think of it like

Sorting photos into labeled folders: vacation, pets, work.

Example

Upload a photo; the system says "This is a dog." Medical imaging: "This X-ray shows pneumonia." Satellite imagery: "This region is urban."

How it actually works

Foundational computer vision task. CNN features (trained on ImageNet or self-supervised) map images to class probabilities. Multi-class (one label per image) vs. multi-label (multiple labels). Transfer learning from pretrained models dominates because labeled data is expensive. Vision transformers and modern CNNs (EfficientNet, ResNet) achieve superhuman accuracy on benchmarks but still fail on distribution shift, adversarial perturbations, and rare classes.

For product teams

Drives content moderation, product tagging, medical diagnosis, and countless automated decisions.

For engineers

Benchmark: ImageNet (1000 classes). Metric: top-1 and top-5 accuracy. Modern models: Vision Transformer, ConvNeXt, ResNet-50.

Read anything AI without the jargon

Look up any term in plain English, or save terms as you read with the free Chrome extension.

Open DecoderAdd to Chrome