When i use face_alignment.FaceAlignment(face_alignment.LandmarksType.THREE_D, flip_input=False), Why the key point shape of the last picture is (1, 1, 68, 3), and the shape of other images are (1, 1, 68, 2)

When i use face_alignment.FaceAlignment(face_alignment.LandmarksType.THREE_D, flip_input=False), Why the key point shape of the last picture is (1, 3, 68, 3), and the shape of other images are (1, 1, 68, 3)

Look at the face_alignment.FaceAlignment class and its get_landmarks method. The issue mentions inconsistent output shapes for 3D landmarks. Check the input processing and how the model handles different image batches or the last image in a sequence. Run a small script to reproduce the shape mismatch with a few test images.

Vom Indexierungsmodell aus dem Issue-Text verfasst.