Inspiration
We wanted to work on a Gym Pose Trainer to use it as a personal trainer as the pandemic is still going on and we have seen that people are not able to go to gyms for doing their exercise. And morover, exercise done with wrong posture can lead to serious health problems like joints pain.
What it does
The system takes live feed from the webcam and then it gives out the pose joints and the coordinates of the joints. The joints are then analyzed using a Random Forest Classifier model which then takes in the joint coordinates and then gives out the detected pose. If the person is doing the exercise in wrong pose, then it would give a red warning. It also counts the number of curls and then displays it in the video feed.
How we built it
We collected the data using Mediapipe library of Python and then stored the coordinates in a CSV file. The CSV file is then used to train a Random Forest Classifier model.
Challenges we ran into
Training the model by collecting the joint coordinates was a bit of challenging for us.
What's next for Gym Pose Trainer
We plan to make a GUI using QT designer and then work on it further to
Log in or sign up for Devpost to join the conversation.