Facial Expression Analysis API

Analyze visible facial landmarks, Action Units, and expression patterns in images or video as time-coded model scores linked to the source media.

What the output means

Labels may use expression categories as a model vocabulary, but they do not prove what a person thinks or feels. Do not use the output to infer intent, truthfulness, mental health, pain, engagement, threat, authenticity, or suitability. Preserve source context and keep consequential interpretation with qualified people.