AI- Personal Fitness Trainer

This project is a AI based fitness training. Recently lot of people facing many health issues due to the lack of physical activity. Not getting enough physical activity can lead to heart disease, even for people who have no other risk factors. It can also increase the likelihood of developing other heart disease risk factors, including obesity, high blood pressure, high blood cholesterol.

In the pandemic social distancing is the big factor, so many workout center are closed. So peoples doing training at home. But there is no a trainer. In the absence of trainer workouts may lead to serious problems.

This project is help you a as trainer at home.It gives the message have doing workout properly and gives how many times you have done your workout.

Software Overview

The software is running on the computer and its written in python.

The main library is opencv and mediapipe.

Step1: Installing Libraries

First of all we are installing the latest version of python on your system.

1.Go to the python website

2.Select the latest version version of python

3.Download the executable installer

4.Run the executable installer

5.Verify python was installed in your system

6. Verify PIP installed in your system

you can check in terminal by

pip
pip --version

Install numpy

pip install numpy

Next we are going to install the opencv

pip install opencv-python

you can check by

python
import cv2
print(cv2.__version__)

Next is Mediapipe

Download the file from Mediapipe official site and install

Step 2: Importing

Import the downloaded library by

import cv2
import mediapipe

Step 3: Landmarks

mpDraw = mp.solutions.drawing_utils
mPose = mp.solutions.pose
pose = mPose.Pose()
while True:
    success, img = cap.read()
    imgRGB = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)
    results = pose.process(imgRGB)
    if results.pose_landmarks:
    mpDraw.draw_landmarks(img, results.pose_landmarks, mPose.POSE_CONNECTIONS)

Step 4: Landmark positions

lmList = []
if results.pose_landmarks:
    mpDraw.draw_landmarks(img, results.pose_landmarks, mPose.POSE_CONNECTIONS)
    for id, lm in enumerate(results.pose_landmarks.landmark):
        h, w, c = img.shape
        cx, cy = int(lm.x*w), int(lm.y*h)
        lmList.append([id, cx, cy])

we can use the points position from the list

print(lmList)

[[0, 278, 335], [1, 287, 306], [2, 299, 306], [3, 310, 305],...........,[32, 287, 1485]]

you can use this image as a point reference

to get the points

x1, y1 = lmList[p1][1:]
>>> 278, 335

Step 5: Angle measurement

By this method we can find the angle by math library

angle = math.degrees(math.atan2(y3-y1,x3-x2)-
                     math.atan2(y1-y2,x1-x2))
print(f'angle:{angle}')

Conclusion:

This project will help to improve the training or workout without any problem.