Inspiration

We wanted to use wrnch's API to recognize sign language with Unity. As this would have required precise recognition of hand movements and a database with words in sign language, we decided to reduce the problem to recognize letters traced in the air by a hand.

What it does

We built a simplified model of two moving dots, each representing a hand, that can trace letters. We were not able to successfully use pose estimation technology to control the two dots, so we decided to control them with the keyboard instead. The tracing can be turned on and off by bringing the two dots into contact. The result is eventually analyzed by an AI to recognize the letter.

How we built it

We used pygame for the user interface and pytesseract for the letter recognition.

Challenges we ran into

We spent a lot of time downloading and installing software, but unfortunately we couldn't get the pose estimation working with wrnch.

Accomplishments that we're proud of

We are proud that, despite all of the problems we encountered and all the time we lost trying to solve them, we still ended up with a final product, even though it is far from what we originally wanted to do.

What we learned

We learned how to use pygame.

Built With

Share this project:

Updates