This project was created out of an interest in applying computer vision and AI to puzzle-solving. The end goal of this project is to make a phone application, capable of reading
pictures taken from the phone. Its main functionality would be to solve puzzles, but I may choose to expand into a more general direction.
I developed the solver logic in Python. Currently, it is able to take a screenshot of a sudoku puzzle from https://sudoku.com,
read the screenshot, solve the sudoku, and automatically inputs the solution back onto the website. The entire process is automated, and is done with the press
of a single button.
The solver is done using a simple AI search technique known as Depth-First Search (DFS). The image reader is done using a technique similar to the k-Nearest Neighbours (kNN)
algorithm. Both solver and reader are using inefficient methods, so it takes approximately 5-10 seconds to solve the sudoku, depending on the complexity.