Inspiration
We were motivated because it was an issue personal to our project lead, Kanchan. She is passionate about skincare and make up, and found the lack of VR resources for trying on products disappointing for people with darker skin tones.
What it does
In its current state, it takes a snapshot of the user's webcam feed, and applies make up filters that they choose for selected areas of their face.
How we built it
We spent half of the time on building the frontend and the other half on the middleware and backend implementations. We used React for the frontend, haskell for the middleware, and python for the backend.
Challenges we ran into
Haskell! It was a challenge sending images between the two programs, particularly because of CORS issues.
Accomplishments that we're proud of
At the eleventh hour, we achieved what we set out to do: apply makeup to the bespoke contours of any user's face. We used every debugging trick we had to discern what the issue was.
What we learned
We learned that we should focus on our anticipated bottlenecks first. We knew that it would be a challenge to communicate between the frontend and the backend, but we vastly underestimated it.
Also, there's something to be said about keeping it simple instead of using less common (though certainly enjoyable) languages.
What's next for Vanity.AI
We have 3 main goals:
- Better match the contours on user's faces
- Utilize machine learning to provide more accurate previews of the make up
- Perform real-time analysis instead of snapshot-based analysis
Built With
- haskell
- react
Log in or sign up for Devpost to join the conversation.