returns a minimal summary of facial features for all detected faces in image.


returns the specified facial features.

Details and Options

  • FacialFeatures, also known as facial characteristics, can return details about detected faces, including position of eyes, nose, etc. as well as global face features, like age, gender, emotion and more.
  • FacialFeatures[{image1, image2,},] can be used to estimate features in multiple images.
  • Facial features can be global facial characteristics or detailed positions of individual facial parts.
  • FacialFeatures[image,"Properties"] returns all available properties and features. The following features are available.
  • Global features:
  • "Age"estimated age
    "Descriptor"feature vector representation of the face
    "Emotion"the most likely facial emotion
    "Gender"estimated gender
    "Landmarks"an association of the facial landmarks
  • Global feature distributions:
  • "AgeWeights"as a WeightedData object
    "EmotionWeights"an association of key-values expr->prob
    "GenderWeights"an association of key-values gender->prob
  • Shape properties:
  • "BoundingBox"bounding Rectangle of the face
    "BoundingBoxArea"area of the bounding box
    "Height"face height
    "Image"a crop of image corresponding to the face
    "Position"position of the face given as {x,y}
    "Width"face width
  • Individual facial landmarks coming back as a list of points:
  • "LeftEyebrowPoints"left eyebrow points
    "LeftEyePoints"left eye points
    "MouthExternalPoints"points of the external part of the mouth
    "MouthInternalPoints"points of the internal part of the mouth
    "NosePoints"points of the nose central line and bridge
    "OutlinePoints"outline of the face
    "RightEyebrowPoints"right eyebrow points
    "RightEyePoints"right eye points
  • Collections of facial landmarks:
  • "EyebrowPoints"an association of left and right eyebrows
    "EyePoints"an association of left and right eyes
    "MouthPoints"an association of the internal and external parts of the mouth
  • By default, the faces are detected using FindFaces and a small padding is added around each bounding box. Use Method{"FaceBoxes"boxes} to specify bounding boxes around faces.
  • Possible values for boxes can be any of the following:
  • Fullthe whole image is assumed to be one face
    {bbox1,}a list of bounding boxes for faces
    fan arbitrary function that returns face bounding boxes
  • FacialFeatures uses machine learning, and its training set and methods may change in different versions of the Wolfram Language, yielding different results.
  • FacialFeatures may download resources that will be stored in your local object store at $LocalBase, and that can be listed using LocalObjects[] and removed using ResourceRemove.


open allclose all

Basic Examples  (2)

Get a summary of the estimated features:

Click for copyable input

Estimate a feature on the faces detected in an image:

Click for copyable input
Click for copyable input

Scope  (11)

Options  (3)

Applications  (4)

See Also

FindFaces  FeatureExtract  Predict  Classify

Introduced in 2018