MediaPipe 0.8 : Colab : MediaPipe ポーズ (翻訳/解説)

翻訳 : (株)クラスキャット セールスインフォメーション
作成日時 : 03/11/2021 (0.8.3)

* 本ページは、MediaPipe の以下のドキュメントを翻訳した上で適宜、補足説明したものです：

• MediaPipe Pose

* サンプルコードの動作確認はしておりますが、必要な場合には適宜、追加改変しています。
* ご自由にリンクを張って頂いてかまいませんが、sales-info@classcat.com までご一報いただけると嬉しいです。

MediaPipe 0.8 : Colab : MediaPipe ポーズ

Python の MediaPipe ポーズ・ソリューション API の使用サンプルです (http://solutions.mediapipe.dev/pose もまた見てください)。

!pip install mediapipe

可視な上半身を持つ人物を持つ任意の画像を Colab にアップロードします。web: https://unsplash.com/photos/4jqfc2vbHJQ と https://unsplash.com/photos/72zsd_fnxYc から 2 つのサンプル画像を取ります。

from google.colab import files

uploaded = files.upload()

import cv2
from google.colab.patches import cv2_imshow

# Read images with OpenCV.
images = {name: cv2.imread(name) for name in uploaded.keys()}

# Preview the images.
for name, image in images.items():
  print(name)   
  cv2_imshow(image)

総ての MediaPipe ソリューション Python API サンプルは mp.solutions 下にあります。

MediaPipe ポーズ・ソリューションについては、mp_holistic = mp.solutions.pose としてこのモジュールにアクセスできます。

初期化の間に static_image_mode と min_detection_confidence のようなパラメータを変更しても良いです。パラメータについてのより多くの情報を得るためには help(mp_holistic.Pose) を実行します。

import mediapipe as mp
mp_pose = mp.solutions.pose

help(mp_pose.Pose)

Help on class Pose in module mediapipe.python.solutions.pose:

class Pose(mediapipe.python.solution_base.SolutionBase)
 |  Pose(static_image_mode=False, upper_body_only=False, smooth_landmarks=True, min_detection_confidence=0.5, min_tracking_confidence=0.5)
 |  
 |  MediaPipe Pose.
 |  
 |  MediaPipe Pose processes an RGB image and returns pose landmarks on the most
 |  prominent person detected.
 |  
 |  Please refer to https://solutions.mediapipe.dev/pose#python-solution-api for
 |  usage examples.
 |  
 |  Method resolution order:
 |      Pose
 |      mediapipe.python.solution_base.SolutionBase
 |      builtins.object
 |  
 |  Methods defined here:
 |  
 |  __init__(self, static_image_mode=False, upper_body_only=False, smooth_landmarks=True, min_detection_confidence=0.5, min_tracking_confidence=0.5)
 |      Initializes a MediaPipe Pose object.
 |      
 |      Args:
 |        static_image_mode: Whether to treat the input images as a batch of static
 |          and possibly unrelated images, or a video stream. See details in
 |          https://solutions.mediapipe.dev/pose#static_image_mode.
 |        upper_body_only: Whether to track the full set of 33 pose landmarks or
 |          only the 25 upper-body pose landmarks. See details in
 |          https://solutions.mediapipe.dev/pose#upper_body_only.
 |        smooth_landmarks: Whether to filter landmarks across different input
 |          images to reduce jitter. See details in
 |          https://solutions.mediapipe.dev/pose#smooth_landmarks.
 |        min_detection_confidence: Minimum confidence value ([0.0, 1.0]) for person
 |          detection to be considered successful. See details in
 |          https://solutions.mediapipe.dev/pose#min_detection_confidence.
 |        min_tracking_confidence: Minimum confidence value ([0.0, 1.0]) for the
 |          pose landmarks to be considered tracked successfully. See details in
 |          https://solutions.mediapipe.dev/pose#min_tracking_confidence.
 |  
 |  process(self, image: numpy.ndarray) -> 
 |      Processes an RGB image and returns the pose landmarks on the most prominent person detected.
 |      
 |      Args:
 |        image: An RGB image represented as a numpy ndarray.
 |      
 |      Raises:
 |        RuntimeError: If the underlying graph throws any error.
 |        ValueError: If the input image is not three channel RGB.
 |      
 |      Returns:
 |        A NamedTuple object with a "pose_landmarks" field that contains the pose
 |        landmarks on the most prominent person detected.
 |  
 |  ----------------------------------------------------------------------
 |  Methods inherited from mediapipe.python.solution_base.SolutionBase:
 |  
 |  __enter__(self)
 |      A "with" statement support.
 |  
 |  __exit__(self, exc_type, exc_val, exc_tb)
 |      Closes all the input sources and the graph.
 |  
 |  close(self) -> None
 |      Closes all the input sources and the graph.
 |  
 |  ----------------------------------------------------------------------
 |  Data descriptors inherited from mediapipe.python.solution_base.SolutionBase:
 |  
 |  __dict__
 |      dictionary for instance variables (if defined)
 |  
 |  __weakref__
 |      list of weak references to the object (if defined)

# Prepare DrawingSpec for drawing the face landmarks later.
mp_drawing = mp.solutions.drawing_utils 
drawing_spec = mp_drawing.DrawingSpec(thickness=1, circle_radius=1)

with mp_pose.Pose(
    static_image_mode=True, min_detection_confidence=0.5) as pose:
  for name, image in images.items():
    # Convert the BGR image to RGB and process it with MediaPipe Pose.
    results = pose.process(cv2.cvtColor(image, cv2.COLOR_BGR2RGB))
    # Print nose landmark.
    image_hight, image_width, _ = image.shape
    if not results.pose_landmarks:
      continue
    print(
      f'Nose coordinates: ('
      f'{results.pose_landmarks.landmark[mp_pose.PoseLandmark.NOSE].x * image_width}, '
      f'{results.pose_landmarks.landmark[mp_pose.PoseLandmark.NOSE].y * image_hight})'
    )

    # Draw pose landmarks.
    print(f'Pose landmarks of {name}:')
    annotated_image = image.copy()
    mp_drawing.draw_landmarks(
        image=annotated_image,
        landmark_list=results.pose_landmarks,
        connections=mp_pose.POSE_CONNECTIONS,
        landmark_drawing_spec=drawing_spec,
        connection_drawing_spec=drawing_spec)
    cv2_imshow(annotated_image)

以上