Fizruk - fitness application for yoga and workout

Fizruk was created in order to practice yoga and workout in real time using computer vision

• Fizruk supports 2 modes: yoga and workout

• There are 3 exercises in each program

  Yoga: T-Pose, Warrior Pose, Tree Pose

  Workout: Arms, Legs, Gym

• Fizruk was created using MediaPipe, OpenCV, PyInstaller.

Example of application operation

Vitrual hands control

The process of pose recognition consists of following steps:

• Recieve video stream using OpenCV.
• Video frames are passed to MediaPipe Pose model that detects pose, adds landmarks (33 landmarks per body) and records their coordinates.
• Landmarks coordinates are extracted, organized and passed to the model for prediction.
• The visualization of model prediction is implemented in the top middle corner of the application screen (see pictures above).
• Detected poses and displayed on top of the screen (see pictures above).

This project was completed in 10 days by:

• https://github.com/AugustVIII
• https://github.com/raulgad
• https://github.com/samot-samoe