GymBro

• 

  cards in the homepage

• 

  feedback and improvements

• 

  analysis screen

• 

  homepage

• 

  check form screen

Inspiration

We thought that since many students go to the gym, having an AI form checker to help them improve their form could be beneficial. Not everyone goes with a buddy or trainer that can always check their form.

What it does

The app analyzes a realtime recorded OR prerecorded video to give you suggestions and constructive criticism about your form. It will also automatically detect what exercise you are doing using a machine learning model.

How we built it

We used React and Tailwind CSS for the frontend. We used Python FastAPI, Google's MediaPipe BlazePose for the joint angles, and a custom Gradient Boosting Classifier from scikit-learn to classify the exercise.

Challenges we ran into

The biggest challenge was with the exercises. Getting the right training and also configurations for the model was difficult at first, but we were able to eventually push through it. Another challenge we faced was with the UI and backend tandem. Trying to get a UI that was smooth and fluid and quick but also a backend that could provide lots of useful was difficult. We were able to find a balance (with the help of Kiro).

Accomplishments that we're proud of

We are proud of creating the machine learning model. This was the first time for both of that we created our own machine learning models. We are also proud of the fact that the app is pretty fast. It doesn't need much time to run the machine learning model and run the feedback scorers/builders.

What we learned

We learned how to create our own machine learning model. We also learned a lot about agentic development and specifically how to use Kiro!

What's next for GymBro

Creating a mobile app, adding more exercise options, and adding more than just form checking.

Built With

• fastapi
• gradient-boosting-classifier
• machine-learning
• mediapipe-blazepose
• python
• react
• scikit-learn