grwm.

Inspiration

Have you ever tried to visualize what clothes and colors fit your vibe and complement you? You might see a color palette, an outfit on social media, or have a certain look in mind, but finding pieces that actually match can mean hours of scrolling through different websites. Even after finding something you like, you still don't really know how it will look on you until you buy it. We wanted to make that process more intuitive by using ML/AI to understand colors that complement you and styles you're looking for; AR let's you see it come to life in 3D before you buy and commit.

What it does

Users start with a face scan, which grwm analyzes using color theory to generate a personalized color palette. We then peruse through our massive clothing collection from multiple vendors that matches those colors and styles that the user may or may not prompt. Finally, grwm lets users virtually try on the clothes in real time, turning a personalized color recommendation into an interactive shopping experience.

How we built it

We built grwm by connecting three core components: a feature-analysis model that turns a face scan into a personalized color palette backed by color theory, the Channel3 API to source matching clothing from multiple vendors (including vendor links, image links, and prices with offers accounted for), and Decart’s Lucy 2.5 model to power real-time virtual try-on. We connected these pieces through a web application that takes the user from their scan, to personalized recommendations, to trying on the clothes virtually. All users are stored via mongoDB.

Challenges we ran into

A big challenge that we ran into was finding a way to connect each individual component of the software. We split it up into 3 subgroups: clothing scraper, full-stack web app, and machine learning model. Finding a way to bridge those three was certainly tricky.

Accomplishments that we're proud of

We’re proud of turning a complex idea into a working end-to-end experience in a short amount of time. We successfully connected our own color-analysis model with Channel3’s multi-vendor product search and Decart’s real-time virtual try-on, creating a smooth process from a user’s face scan to personalized clothing recommendations and actually seeing those clothes on themselves.

What we learned

We learned how to bring multiple AI models and APIs together into one cohesive product. We gained hands-on experience working with real-time AI, API integration, and building an end-to-end application under a tight deadline.

What's next for grwm.

We want to implement a way for there to be a side-by-side view of different computations of outfits that grwm helps the user pick. We would also love to include specified filters, like budget friendly options, while also implementing more of what Channel3's API can give to us like availability and brand filtering, almost feeling like a real retail platform.