Decoder. plain-English AI glossary

Keypoint detection

▲ Rising

Spotting specific landmarks — eyes, joints, edges — in an image so the model knows where important parts are.

Think of it like

Like marking constellations on a star map: the model learns to highlight the brightest points in an image.

Example

A pose estimator locates your shoulders, elbows, and knees so a fitness app can check your form during a workout.

How it actually works

The model outputs coordinates for each landmark, often with confidence scores. Accuracy matters downstream — a missed shoulder throws off everything built on it. Modern approaches use heatmaps or regression heads to predict positions end-to-end.

For product teams

Unlocks use cases that need to understand *where* things are, not just *what* they are — from fitness tracking to gesture control.

For engineers

Output is typically a list of (x, y) coordinates or a heatmap per keypoint; training uses L2 loss or MSE on predicted positions.

Related

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