Inspiration
Gen Z is very conscious about what they wear. We wanted to create a social media application for the younger generation to share their different styles (fits) with other users and receive feedback
Although nowadays it feels like social media is being criminalized, we believe deep down that something like fashion can bring people together :)
What it does
Fit Check's main feature is a "for you" styled page where users can scroll through different fits and rate them out of five. Additionally, we utilize Artificial Intelligence (AI) to analyze the "fits" posted and find similar products online and provide a link to the store.
We were short on time for creating a different "explore" page vs. "for you" page where data based on what a user rated highly would impact the order of fits whereas the "explore" page would just be all fits
Our app also contains an account page where one may see all their "fits" in the past as well
Finally, our app has a secure login and logout feature
How we built it
Our for you page recommendation system uses two AI API services. Firstly, OpenAI image classification is used to determine the style of clothing a person in a picture is wearing. It is able to recognize a variety including tops, bottoms, dresses, watches, etc. Then, we feed the data into Google where we get results for a store with a similar product
We were actively working on a system where based on what users rate highly in their explore page, we would make to use decision. We would make decisions using how similar a user's preference matrix was to a product's matrix and rank products on the for you page that way.
On the backend, we used Flask. Flask was nice to use because a lot of the API work and ML stuff is easy to do in Python compared to other languages
Challenges we ran into
Sometimes the APIs just would not work and we had to figure out why. We had a lot of trouble integrating the front and backend. It definitely became a larger lesson on prepping tech stacks.
Accomplishments that we're proud of
A majority learned to implement React Typescript and the rest learned to implment more APIs into their backend.
What we learned
We learned to deal with APIs more. How to get API tokens with Google, OpenAI. We initially used Google instead of OpenAI and it was much worse, so we switched over
What's next for FitCheck
The recommender system for the for you page will continue to be fine tuned and the UI for the project will also be touched up and leaning more towards react.
Log in or sign up for Devpost to join the conversation.