pose_estimator.py

import mediapipe as mp
from mediapipe.tasks import python
import cv2
import numpy as np
from mediapipe import solutions
from mediapipe.framework.formats import landmark_pb2

file_name='IMG_2149.mov'
file_name = file_name
model_path = 'pose_landmarker_full.task'

# https://developers.google.com/mediapipe/solutions/vision/pose_landmarker/python#video
options = python.vision.PoseLandmarkerOptions(
 base_options=python.BaseOptions(model_asset_path=model_path),
 running_mode=python.vision.RunningMode.VIDEO)

def draw_landmarks_on_image(rgb_image, detection_result):
 pose_landmarks_list = detection_result.pose_landmarks
 annotated_image = np.copy(rgb_image)

 for idx in range(len(pose_landmarks_list)):
 pose_landmarks = pose_landmarks_list[idx]

 # Draw the pose landmarks.
 pose_landmarks_proto = landmark_pb2.NormalizedLandmarkList()
 pose_landmarks_proto.landmark.extend([
 landmark_pb2.NormalizedLandmark(x=landmark.x, y=landmark.y, z=landmark.z) for landmark in pose_landmarks
 ])
 solutions.drawing_utils.draw_landmarks(
 annotated_image,
 pose_landmarks_proto,
 solutions.pose.POSE_CONNECTIONS,
 solutions.drawing_styles.get_default_pose_landmarks_style())
 return annotated_image

with python.vision.PoseLandmarker.create_from_options(options) as landmarker:
 cap = cv2.VideoCapture(file_name)
 fps = cap.get(cv2.CAP_PROP_FPS)
 calc_timestamps = [0.0]

 if (cap.isOpened()== False): 
 print("Error opening video stream or file")

 while(cap.isOpened()):
 ret, frame = cap.read()
 if ret == True: 
 mp_image = mp.Image(image_format=mp.ImageFormat.SRGB, data=frame)
 calc_timestamps.append(int(calc_timestamps[-1] + 1000/fps))
 detection_result = landmarker.detect_for_video(mp_image, calc_timestamps[-1])
 
 annotated_image = draw_landmarks_on_image(frame, detection_result)
 cv2.imshow('Frame',annotated_image)

 # Press Q on keyboard to exit
 if cv2.waitKey(25) & 0xFF == ord('q'):
 break
 else: 
 break
 
 cap.release()
 cv2.destroyAllWindows()