Liza was immensely frustrated. The new clothes she ordered do not fit her well. Again. If the shoulders fit, the bottom flared out a little too much. She was tired of paying for the returns and just losing money through the entire process.

What it does

An advert in Washington Times caught her eye. A customizable FittingRoom AR? A small ray of hope penetrated the walls of frustration that she had locked herself in. A personal recommender that gauranteed the best possible fit for her. She decided one last time.

How we built it

We first got data on user purchase history for clothes. Then we used it to create a recommender system in python that recommends products for any given user. The outputs were handed to the AR app which is built using unity. The AR app used 3D models to generate fits.

Challenges we ran into

The majority of the trouble figuring out how to get the AR part working since no one in the team had developed AR before.We had models of humanoids and models of clothes. However, they were of different sizes and scaling them down was difficult.

Accomplishments that we're proud of

“Hmmm..” She thought. She liked the user interface. It was easy to use and the model that the app generated was so accurate. It represented her measurements perfectly. She was amazed. The recommendations were so fashionable.and looked great on her frame. She was able to see what she could see in a store fitting room in her own home.

What we learned

What's next for FittingRoom AR

Integrating inventories from worldwide retail stores with the existing online store to tap the entire digital market

Built With

  • arcore
  • google-sceneform
  • java
Share this project: