Reference · Glossary

Computer vision

Last updated

AI that **interprets images and video** — detect objects, read text (OCR), segment regions, estimate poses, or classify scenes.

When to use

Quality inspection, document scanning, accessibility alt-text, security cameras, medical imaging assist (with human oversight).

When not to

When a text-only LLM is enough — don't pipe screenshots through vision if plain text metadata already exists.

Example

A warehouse camera flags pallets missing safety wrap; a human reviewer confirms before stopping the line.