Facial Landmarks Result Scheme

Result scheme for the Facial Landmarks algorithm in Face Metrics 2.0, which returns a vectorized mask of 53 predefined face points.

Facial landmarks are a set of face points that allow the marking of facial features, such as the nose, mouth, eyes, etc. Haut.AI provides 53 predefined landmark points:

facial landmarks

algorithm_tech_name

facial_landmarks

area_results

area_results is returned as an empty list

masks_restored

The algorithm returns a vectorized mask of the facial landmarks for a restored image

• "mask_type": "point_mask"

• features

  • geometry

    • "type": "MultiPoint"

    • "coordinates" list of (x,y,z) landmarks coordinates. The first element of the list corresponds to the point 1 from the image above, the second corresponds to the point 2, etc.

    NB! Note that now z (depth) coordinate returns 0 for all points. This value is reserved for a future update. Use (x,y) coordinates only.

  See more on Masks here

masks_original

The algorithm returns a vectorized mask of the facial landmarks for an original image

• "mask_type": "point_mask"

• features

  • geometry

    • "type": "MultiPoint"

  See more on Masks here

Example (JSON)

landmarks.json

19KB

Open

landmarks.json

---

Not a Haut.AI client yet?

Book a demo Platform overview on haut.ai