Video measurements

The image upload and video realtime endpoints return two arrays of scores for each measured face: expressions and descriptions. Both are null for a face beyond the limit on measured faces. Scores explains how to read them. The video and audio endpoints use different expression names.

{
  "expressions": [
    {
      "name": "amusement",
      "probability": 0.7124
    },
    {
      "name": "joy",
      "probability": 0.5308
    }
  ],
  "descriptions": [
    {
      "name": "smile",
      "probability": 0.8419
    },
    {
      "name": "grin",
      "probability": 0.3172
    }
  ]
}

Expressions

Emotions and mental states a person viewing the face would name, such as amusement and concentration. 48 names.

Descriptions

Facial actions visible on the face, such as smile or jaw drop. 27 names. Each probability is the chance that the action is present. See Scores.

Next steps