BASIC POSE DETECTION AND TRACKING USING PYTHON

    High fidelity human body pose tracking, inferring up to 33 3D full-body landmarks from RGB video frames.

The 33 full-body landmarks will be like, 

And, The output will be like this, 

Packages to install : 

1. pip install opencv-python.
2. pip install mediapipe.

Program : 

# Pose Estimation with Python
import cv2
import mediapipe as mp
import time

# Capturing video from the web cam
video = cv2.VideoCapture("videos/running.mp4")

# Creating object for pose detection
myPose = mp.solutions.pose
pose = myPose.Pose()

mpDraw = mp.solutions.drawing_utils

currentTime = 0
previousTime = 0

while True:
    success, img = video.read()
    # Converting BGR image to RGB image
    RGBimg = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)
    img = cv2.resize(img, None, fx=0.15, fy=0.15)

    # Getting landmark points and storing in poseLandmarks variable
    result = pose.process(RGBimg)
    poseLandmarks = result.pose_landmarks

    # This 'if' block will run when it's having the poseLandmark points
    if poseLandmarks:
        # Accessing each of the 33 landmarks
        """
        for index, poseLandmark in enumerate(poseLandmarks.landmark):
            w, h, c = img.shape
            cx, cy = int(poseLandmark.x * h), int(poseLandmark.y * w)
            if index == 11:
                cv2.circle(img, (cx, cy), 10, (255, 0, 0), cv2.FILLED)
        """
        # Drawing and connecting the points
        mpDraw.draw_landmarks(img, poseLandmarks, myPose.POSE_CONNECTIONS)

    # Calculating frames per second
    currentTime = time.time()
    fps = 1 / (currentTime - previousTime)
    previousTime = currentTime

    # This will show the fps value on screen
    cv2.putText(img, str(int(fps)), (30, 70), cv2.FONT_HERSHEY_COMPLEX, 2, (186, 4, 22), 5)

    cv2.imshow("Pose Detection", img)
    key = cv2.waitKey(1)
    if key == 13:
        exit()