Inspiration

The workout industry has seen an increase in exposure within the past couple of years and continues to grow. Specifically, there were 57 million health club members in 2016, meaning 3.6% more than in 2015. However, the increase in gym memberships does not account for how often people actually go to the gym. According to some more statistics, gym members go to the gym on average two times a week and about 67% of people never even use their gym memberships. With gym memberships costing on average $58 per month, about $39 is wasted from underutilization; consequently, a total of $2.2 billion dollars go down the drain. Common reasons that cause these statistics include: people not having the time, people are lazy, and no one will know when we skip a workout.

What it does

Because we live in a society that sees an increase in incorporating games into normal life tasks, we thought it would be a perfect opportunity to use the concept in the fitness environment. Socially, users can connect to and work out with anyone from around the world while the gaming and competitive environment promotes fun and challenge; economically, people don't have to pay for a gym membership to get a work out done or pay gas to travel to the gym; and responsibly, public exposure of people's statistics provides accountability.

How we built it

We created an iOS mobile application using Swift 4. On top of that, we have AVFoundation for QR code scanning, as well as CoreML with a convolutional LSTM neural network trained for physical activity. This information is passed to the realtime NoSQL backend database, which is then displayed by the Angular 4 web-app.

Challenges we ran into

Half of our team do not come from a computer science background, and thus it was a great learning experience for everyone. We also ran into various issues with the machine learning, as well as overall analysis of the physical activity.

Accomplishments that we're proud of

We managed to create a working project with an extremely accurate fitness tracker.

What we learned

Machine learning is difficult.

What's next for FitBuddy

We're really passionate about FitBuddy and plan on continuing to develop and improve it. Some functionality that we have in mind include: unlocking trophies or nicknames after accomplishing certain achievements, winning streaks, leaderboards for different regions on adjustable scales, expanding exercises into weighted movements and cardio, a matching system to provide fair competition, and many more.

Share this project:
×

Updates