SuitUp

We developed this algorithm to determine the similarity between clothes and to recommend sets of outfits to users.

Comment

Inspiration

One of our team members noticed friends that would comment on his/her style choice frequently. So, we decided to create an algorithm to suggest outfits to people who do not have the best style choices.

What it does

This project determines the similarity between clothes and to recommend sets of outfits to users. The greatest cosine value between pair meant they were similar. We used AI to enhance the recommendation output based on the users’ preferences.

How we built it

We web scraped Tommy Hilfiger to retrieve a small sample set of data and stored that data in a MongoDB database. We assigned values to each items’ attributes, stored the values in a vector, and calculated the cosine similarity between each pair of clothes based on each items' attribute vector.

Challenges we ran into

We faced challenges when we were assigning values to attributes and determining each item's vector. In addition, we had difficulty creating a remote server to run our algorithm. So we decided to keep the algorithm local, and deploy later.

Accomplishments that we're proud of

Successfully using MongoDB database. Algorithm gives very similar outfit recommendation if we inputted an output from our initial input.

What we learned

We learned a lot about recommendation systems, and the two different types of them. We learned about machine learning as well as how to use the MongoDB framework.

What's next for SuitUp

We hope to expand this project by incorporating more clothing brands. We also hope to deploy our project on a server to run remotely.

Share this project:

Updates