59. IntelliFit

Inspiration

After witnessing the abundance of unhealthy choices presented to us in daily life, we had the desire to create a dedicated fitness platform where people could have personalized workouts to make it easier to stay healthy.

What it does

Using a user interface, a user can create an account to track their progress over time. The user also has the ablitiy to select between different workouts and craft custom workouts to suit their fitness needs. Once a workout is chosen, the user is guided through their workout with pose estimations layered on top of their figure at relevant positions depending on the current exercise. A counter keeps track of the number of repetitions of the exercise and the accuracy of the form of the individual is measured.

How we built it

The backend was written using python using mediapipe for the purpose of pose estimation. The front end was written in React in order to develop dynamic web pages, along with the use of Firebase to store user information.

Challenges we ran into

It was very difficult to integrate the frontend and backend together so that images of the workout could be sent to the model and the annotated image could be sent back. We also struggled with calibrating the model to work with a variety of exercises as each one had to be trained separately.

Accomplishments that we're proud of

We are proud that despite the many challenges we faced, we kept persevering to try to overcome them.

What we learned

We learned to never underestimate the task of connecting the frontend and backend together as this is an issue that can be very hard to solve.

What's next for IntelliFit

The model could be improved to produce better pose estimations. Additionally, more exercises could be added so more complex workouts could be created. We can also work on improving communication between front and back end for a more efficient program.