Dressr

How do you know what to wear? There's an app for that.

Comment

Inspiration

Our visionary, Arham, has trouble picking out what clothes to wear. He's going to college next year, so his mom won't be around to pick out clothes for him. So we asked the question: can we write a smart algorithm to pick out good clothes, and save Arham's mother the pain of daily wardrobe planning? And after 36 hours of hacking, we have the answer, in the form of Dressr.

What it does

Tell Dressr what's in your closet - and it'll have you take a look at some outfits. Upvote and downvote the outfits to tell Dressr what you like and dislike, to train the outfit picker on what to do and what not to do.

How we built it

We set up a pretty cool data flow to make this app work. We store the data on the user's closet in a Firebase database, and access it through their node.js package. Eashwar wrote a RESTful API using express.js, so both the front end (HTML/CSS) and the outfit generation algorithm (python) could access it with ease. Lexie's outfit generation algorithm uses the flask framework for the same cross-language data flow. Arham's front end was built with a custom CSS library, and uses an svg fontpack of clothing line drawings to display both closet items and outfits.

Challenges we ran into

We initially pursued using an AI with machine learning to generate the outifts, but this ended up being unreasonably hard to accomplish in the 36 hour timespan. Instead, we wrote a smart algorithm that essentially runs a control loop, generating outfits based on what the user upvotes and downvotes.

Accomplishments that we're proud of

The algorithm for generating outfits has a control loop, which allows it to take user feedback. We were able to make a cross-language data flow so that each developer could develop in the language that best suited their needs. Our UI uses a custom CSS library, developed by Arham: ajcss, and is fully responsive: the moment you begin to type in a color or article of clothing, it predicts what you're writing.

What we learned

Artifical Intelligence is very, very hard to implement, especially in 36 hours. We also learned a lot about using flask and express.js as web frameworks.

What's next for Dressr

A user system so each person has their own closet. This will enable us to run online securely. Some AI integration Adhesive RFID tags that can be applied to clothing that would immediately register that article with Dressr.

Built With

Share this project:

Updates