python - mediapipe 实时 3D 坐标打印

Question

我想打印 mediapipe 3D 地标的实时 3D 图。 我能够从视频文件中实时获取如下所示的整体信息: 现在我想绘制实时 3D 图，如下所示:

我用来绘制实时整体的代码如下所示。我如何能够仅从整体部分绘制实时图？

import cv2
import mediapipe as mp
import urllib.request
import numpy as np
import pickle
import matplotlib as mpl
import matplotlib.pyplot as plt
from matplotlib import animation
import PyQt5
from PIL import Image
from IPython.display import Video
import nb_helpers

mp_drawing = mp.solutions.drawing_utils
mp_holistic = mp.solutions.holistic 

file = 'walking.mp4'
cap = cv2.VideoCapture(file)

# Initiate holistic model
with mp_holistic.Holistic(min_detection_confidence=0.5, min_tracking_confidence=0.5) as holistic:
    
    while cap.isOpened():
        ret, frame = cap.read()
        
        # Recolor Feed
        image = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)
        image.flags.writeable = False        
        
        # Make Detections
        results = holistic.process(image)
        # print(results.face_landmarks)
        
        # face_landmarks, pose_landmarks, left_hand_landmarks, right_hand_landmarks
        
        # Recolor image back to BGR for rendering
        image.flags.writeable = True   
        image = cv2.cvtColor(image, cv2.COLOR_RGB2BGR)
        
        # 1. Draw face landmarks
        mp_drawing.draw_landmarks(image, results.face_landmarks, mp_holistic.FACEMESH_TESSELATION, 
                                 mp_drawing.DrawingSpec(color=(80,110,10), thickness=1, circle_radius=1),
                                 mp_drawing.DrawingSpec(color=(80,256,121), thickness=1, circle_radius=1)
                                 )
        
        # 2. Right hand
        mp_drawing.draw_landmarks(image, results.right_hand_landmarks, mp_holistic.HAND_CONNECTIONS, 
                                 mp_drawing.DrawingSpec(color=(80,22,10), thickness=2, circle_radius=4),
                                 mp_drawing.DrawingSpec(color=(80,44,121), thickness=2, circle_radius=2)
                                 )

        # 3. Left Hand
        mp_drawing.draw_landmarks(image, results.left_hand_landmarks, mp_holistic.HAND_CONNECTIONS, 
                                 mp_drawing.DrawingSpec(color=(121,22,76), thickness=2, circle_radius=4),
                                 mp_drawing.DrawingSpec(color=(121,44,250), thickness=2, circle_radius=2)
                                 )

        # 4. Pose Detections
        mp_drawing.draw_landmarks(image, results.pose_landmarks, mp_holistic.POSE_CONNECTIONS, 
                                 mp_drawing.DrawingSpec(color=(245,117,66), thickness=2, circle_radius=4),
                                 mp_drawing.DrawingSpec(color=(245,66,230), thickness=2, circle_radius=2)
                                 )
                        
        cv2.imshow('Raw Webcam Feed', image)

        if cv2.waitKey(10) & 0xFF == ord('q'):
            break

cap.release()
cv2.destroyAllWindows()

如果有任何帮助，我将不胜感激!

问候!

Answer 1

您可以尝试下面的代码，这是我从https://github.com/Kazuhito00/mediapipe-python-sample/blob/main/sample_pose.py获得的

以下是您可以包含的用于绘制实时 3D 地标的函数

def plot_world_landmarks(
    plt,
    ax,
    landmarks,
    visibility_th=0.5,
):
    landmark_point = []

    for index, landmark in enumerate(landmarks.landmark):
        landmark_point.append(
            [landmark.visibility, (landmark.x, landmark.y, landmark.z)])

    face_index_list = [0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10]
    right_arm_index_list = [11, 13, 15, 17, 19, 21]
    left_arm_index_list = [12, 14, 16, 18, 20, 22]
    right_body_side_index_list = [11, 23, 25, 27, 29, 31]
    left_body_side_index_list = [12, 24, 26, 28, 30, 32]
    shoulder_index_list = [11, 12]
    waist_index_list = [23, 24]

    # 顔
    face_x, face_y, face_z = [], [], []
    for index in face_index_list:
        point = landmark_point[index][1]
        face_x.append(point[0])
        face_y.append(point[2])
        face_z.append(point[1] * (-1))

    # 右腕
    right_arm_x, right_arm_y, right_arm_z = [], [], []
    for index in right_arm_index_list:
        point = landmark_point[index][1]
        right_arm_x.append(point[0])
        right_arm_y.append(point[2])
        right_arm_z.append(point[1] * (-1))

    # 左腕
    left_arm_x, left_arm_y, left_arm_z = [], [], []
    for index in left_arm_index_list:
        point = landmark_point[index][1]
        left_arm_x.append(point[0])
        left_arm_y.append(point[2])
        left_arm_z.append(point[1] * (-1))

    # 右半身
    right_body_side_x, right_body_side_y, right_body_side_z = [], [], []
    for index in right_body_side_index_list:
        point = landmark_point[index][1]
        right_body_side_x.append(point[0])
        right_body_side_y.append(point[2])
        right_body_side_z.append(point[1] * (-1))

    # 左半身
    left_body_side_x, left_body_side_y, left_body_side_z = [], [], []
    for index in left_body_side_index_list:
        point = landmark_point[index][1]
        left_body_side_x.append(point[0])
        left_body_side_y.append(point[2])
        left_body_side_z.append(point[1] * (-1))

    # 肩
    shoulder_x, shoulder_y, shoulder_z = [], [], []
    for index in shoulder_index_list:
        point = landmark_point[index][1]
        shoulder_x.append(point[0])
        shoulder_y.append(point[2])
        shoulder_z.append(point[1] * (-1))

    # 腰
    waist_x, waist_y, waist_z = [], [], []
    for index in waist_index_list:
        point = landmark_point[index][1]
        waist_x.append(point[0])
        waist_y.append(point[2])
        waist_z.append(point[1] * (-1))
            
    ax.cla()
    ax.set_xlim3d(-1, 1)
    ax.set_ylim3d(-1, 1)
    ax.set_zlim3d(-1, 1)

    ax.scatter(face_x, face_y, face_z)
    ax.plot(right_arm_x, right_arm_y, right_arm_z)
    ax.plot(left_arm_x, left_arm_y, left_arm_z)
    ax.plot(right_body_side_x, right_body_side_y, right_body_side_z)
    ax.plot(left_body_side_x, left_body_side_y, left_body_side_z)
    ax.plot(shoulder_x, shoulder_y, shoulder_z)
    ax.plot(waist_x, waist_y, waist_z)
    
    plt.pause(.001)

    return