Inspiration
We were buying clothes for ourselves online, but we didn't know if it would fit us so we decided we can just return it after trying it on. Not only did we find this very inefficient, we also found out later that this system actually contributes to climate change.
What it does
Our app has an AR try on feature, allowing users to see how they look like wearing the clothes they want to buy. We also coded an ML model that recommends you clothes based on what you like.
How we built it
For the AR model, we used ARKit and RealityKit. For the ML model, we used CreateML and CoreML to create the model. We created a few parameters that determined what to recommend. For example, if I chose a black crewneck shirt, the model would recommend black and/or crewneck shirts. Other parameters are long/short sleeve and design.
Challenges we ran into
When making the AR model, we had to use motion capture with ARKit and RealityKit to create a skeleton of the body. Then, we had to overlay the images of the shirt onto the skeleton model to show how it would look like on the user. When making the ML model, we ran into a few issues with rendering the images and getting the recommendations to work because it wasn't recommending the correct things.
Accomplishments that we're proud of
We're proud of created a fully working AR motion capture model and also creating and ML model that can accurately recommend what clothes you would like based on your preferences.
What we learned
We learned how to use Motion Capture with SwiftUI and also how to create a machine learning model with CreateML and CoreML.
What's next for Fashion Killa
We want to further improve the AR model to depict the clothes better. We also what to make an even more accurate ML model and expand the existing dataset.
Log in or sign up for Devpost to join the conversation.