MatchMyPose

Inspiration

Exercising in quarantine during the pandemic has become a very tough thing. To increase the motivation level maintain the fitness level and to maintain form and prevent injuries during exercising, We built this app.

What it does

This is an ML based app which compares the instructors video and the users video, to tell the differences between the two and give feedback telling the Accuracy percentage and the parts with the errors.

How I built it

We used Flutter to build the mobile app, We used a firebase database for the authentication and storing the realtime database of workouts. We also built the machine learning model based on pose detection.

Challenges I ran into

After building both, the Flutter app with a backend and the machine learning model, we could not figure out how do we use the machine learning model in the flutter app even after 7 hours of research. We had to hard-code the evaluation page due to lack of time

Accomplishments that I'm proud of

I am very proud of the flutter app, its UI and its firebase databse. I am also proud of the machine learning model made by Gurusha as it its really amazing!

What I learned

We should focus on the main things and the backend of the app first and make it work before beautifying it Managing time, learning from each others' failures and successes

Problems

We could not figure out how to run the ML code from the flutter app, so we had to hardcode the evaluation screen for now.

What's next for MatchMyPose

A proper implementation for the Machine Learning model inside the app and giving a proper evaluation report, instead of hardcoding it, A better UI and a blog-type place for articles, A separation for trainers and students, payment options for trainers and much more.

Built With

• android-studio
• dart
• firebase
• flutter
• google-cloud-services
• javascript
• python
• vs-code