We wanted to learn more about image processing, and we all enjoy the fun game of Sudoku
What it does
It allows you to upload a picture of a Sudoku puzzle and it will solve it, and display the solution in an easy-to-see grid
How we built it
A python backtracking algorithm that solves the actual puzzle. We also used neural networks to recognize handwriting. Finally, we used react to build a frontend.
Challenges we ran into
Unfamiliarity with CNNs was a challenge, as well as never having used OpenCV with Python
Accomplishments that we're proud of
Despite being relatively unfamiliar with these technologies, we managed to get an application that works decently well. We are also proud of our number recognition which we made ourselves using TensorFlow
What we learned
We learned a lot about image processing and OpenCV. Some people on the team have never written Python before, so that was also a good learning opportunity
What's next for sudoku-solver
Perhaps an application that automatically sends a picture you take on your phone to the solver API. If we really want to get fancy, maybe some AR stuff.