Stuck at home during the pandemic, behind a camera, we realized we had all lost a sense of our style: our confidence. We felt more out of touch with the trends even though we spent hours scrolling through the web. There was no website or app or any coordinated place to look for to see what's trending in fashion. In need of a fashion guru, we decided that we should just build it ourselves.
What it does
The app uses an ML model that learns from the user's current closet to recommend an outfit. It has 3 useful features: recommending a daily outfit, suggesting an article of clothing to buy based on the rest of the fit, and finally rating the user's outfit on trendiness.
How we built it
We used Figma to design the basic UI and communicated tasks through JIRA. To build the frontend UI we used React Native because it was the most well-known and has a lot of library support. To build the backend, we used google cloud server and python to run the ML server.
Challenges we ran into
Building the ML model was the most difficult task. Additionally, since we decided to make the React project in Typescript, following best practices in the industry, we ran into errors when importing libraries. This debugging caused a lot of delays in development.
Accomplishments that we're proud of
Figuring out how to code an ML model with almost no prior knowledge. And pulling an all-nighter ;)
What we learned
We learned how to make decisions on what technology to choose based on given deadlines. Additionally, we learned new shortcuts for programming.
What's next for FITTR
We need to improve FITTR's ML model to incorporate a feature that suggests what clothes to buy based on a given outfit. This can allow this platform to grow economically and gain traction among more users.