[python] face landmarks, keypoints, roi

Face Mesh Landmark

FACEMESH_NUM_LANDMARKS = 468

FACEMESH_NUM_LANDMARKS_WITH_IRISES = 478

Face Mesh Return

multi_face_landmarks (namedTuple)

468개의 x, y, z

x, y 는 [0.0] ~ [1.0] 으로 normalized 된 값이고 z는 depth를 나타냄. 작은 값일 수록 카메라와 가까움을 나타냄.

 

https://google.github.io/mediapipe/solutions/face_mesh.html#multi_face_landmarks

Face Mesh

 

Face Mesh Keypoints

photo:

https://github.com/google/mediapipe/blob/a908d668c730da128dfa8d9f6bd25d519d006692/mediapipe/modules/face_geometry/data/canonical_face_model_uv_visualization.png

names:

https://github.com/tensorflow/tfjs-models/blob/838611c02f51159afdd77469ce67f0e26b7bbb23/face-landmarks-detection/src/mediapipe-facemesh/keypoints.ts

GitHub - tensorflow/tfjs-models: Pretrained models for TensorFlow.js

 

ROI

.cc

mediapipe::NormalizedRect roi

x_center, y_center, width, height, rotation

ref: mediapipe/calculators/tensor/image_to_tensor_calculator_test.cc

.pbtxt

input_stream: "NORM_RECT:roi"

 

Face Detection Keypoints

class FaceKeyPoint(enum.IntEnum):
  """The enum type of the six face detection key points."""
  RIGHT_EYE = 0
  LEFT_EYE = 1
  NOSE_TIP = 2
  MOUTH_CENTER = 3
  RIGHT_EAR_TRAGION = 4
  LEFT_EAR_TRAGION = 5

Face Detection Returns

Detections (namedTuple)

bounding box(xmin, ymin, width, height), 6개의 keypoints

 

 

References

https://github.com/google/mediapipe/issues/1615

Definitions for 468 landmark points of a face · Issue #1615 · google/mediapipe