AI Glossary

Data labeling

Also called data annotation

What does “Data labeling” mean?

Data labeling: Adding correct answers, tags or ratings to raw data so it can be used for supervised training or evaluation. People, models or both can do the work, and label quality has a strong effect on how well the trained model performs.

Data labeling visual guide: Adding correct answers, tags or ratings to raw data so it can be used for supervised training or evaluation. People, models or both can do the work, and label quality has a strong effect on how well the trained model performs.
Visual guide Open full size (new tab)

What is an example of “Data labeling”?

Drawing boxes around pedestrians and cars in street photos is labeling work that helps train perception systems for self-driving vehicles.

Keep learning

See also

Supervised learningTermSynthetic dataTermReinforcement learning from human feedbackTermTraining dataTerm

Same topic

Direct preference optimizationTermCurriculum learningTermEpochTermComputeTermFine-tuningTerm