Move2Note

Do you find gym workouts to be monotonous and repetitive? Using Move2Note, follow along to a song and turn your workout into a song by mapping exercises to musical notes!

Comment

Inspiration

Common exercise workout routines revolve around a sets of repetitions, such as 3 sets of 8. However, we found this to be rather boring and unfulfilling. We wanted routines to be dynamic and exciting, such that the user could have fun whilst working out and feel accomplished at the end of the workout routine.

What it does

Move2Note is a webapp that maps notes in a song to exercises. For example, a C note is a Sit-up. Following this logic, a song can be converted into a sequence of exercises. Users can then follow the instructions on the webapp to complete the workout routine. Users can upload a video of their routine to Move2Note and their chosen song will be played back to them! Ideally, this gives users a sense of achievement.

How we built it

Following online videos and guides on OpenCV and Google Mediapipe, we built a base foundation for which pose estimation could be reliably extracted from videos. We used a combination of self-made and open source videos to train and test our model. With the help of a coding agent, we could seamlessly iron out bugs and speed up development. Our frontend contains HTML, CSS and JS code. Our backend is built using Python running FastAPI, served via Uvicorn.

Challenges we ran into

We took a while to set up the frontend and backend as we had little background in full stack development. There was a need to rely on online materials and AI to configure it properly as we wanted it.

Accomplishments that we're proud of

Coming into this hackathon, we did not expect to get a fleshed-out prototype within 24 hours. We were surprised to see how helpful AI could be in Software Engineering as a learning aid and coding tool. We are proud to have completed a mini-project that was merely an far-fetched idea prior to the start of the hackathon.

What we learned

We developed a greater understanding on how to use and call APIs, as well as use frameworks such as OpenCV and Mediapipe.

What's next for Move2Note

With more time, a variety of other exercises such as bicep curls or deadlifts could be added to increase the number of notes available for users to use. Additionally, more variable data (lighting, angle, person) could be used in the training of the model to ensure more reliable results.

Built With

Share this project:

Updates