These changes allow the user to also return the face bounding box, when calling get_landmarks_from_image or get_landmarks_from_batch, I think it doesn't break existent code and enables a second useful output.
The option to enable this feature is passed as the last argument of FaceAlignment constructor (return_face_results). By having it disabled by default, the return values of the two get_landmarks_* remain as they currently are (only landmarks), therefore doesn't break existent code that expects single output. Also because of the default value in the constructor, it doesn't object creation with less or named arguments.
Hi,
These changes allow the user to also return the face bounding box, when calling
get_landmarks_from_image
orget_landmarks_from_batch
, I think it doesn't break existent code and enables a second useful output.The option to enable this feature is passed as the last argument of
FaceAlignment
constructor (return_face_results
). By having it disabled by default, the return values of the twoget_landmarks_*
remain as they currently are (only landmarks), therefore doesn't break existent code that expects single output. Also because of the default value in the constructor, it doesn't object creation with less or named arguments.