Create a face
Adds a user-drawn face box to an asset, for a face the detector missed. To remove a face detection instead, use delete_face; to introduce a brand-new identity first, use create_person.
Body
ID of the asset (with asset_ prefix) to draw the face box on. Get IDs from list_assets / search_assets. The asset must belong to the target library.
Where the face is, as a box in display-space pixels matching the asset's reported width/height. The box must fit inside those dimensions.
Optional person ID (with person_ prefix) to assign this face to at creation. Omit to leave it unassigned; assign it later via update_face. Get IDs from list_people; use create_person first if the identity doesn't exist yet.
Library to create the face in. Optional if the user has a single library; required when they have multiple.
Response
Successful Response
Represents a detected face in an asset with facial recognition data.
Unique face identifier with 'face_' prefix
ID of the asset containing this face
Face location as {x, y, w, h} coordinates in pixels
How this face was added: 'automatic' for detector-found faces, 'manual' for user-drawn face boxes.
automatic, manual When this face was detected and recorded
When this face record was last updated
ID of the person this face belongs to (if identified)
Detector confidence on a 0-1 scale; higher is more confident among faces detected under the same configuration (values are not comparable across detector generations). Null on legacy faces without a stored score and on manually added faces.
For video files, timestamp in milliseconds when face appears
Asset variants for this face: 'thumbnail' with face crop
Cluster-assignment diagnostics for this face. Populated only when include=cluster_assignment is requested on the faces endpoint; null otherwise. See ClusterAssignmentResponse for the shape.