The project consists of 3 parts:
- Computer Vision - To detect and extract a sudoku from a camera input.
- Deep Learning - To predict whether each cell contains a number (1 to 9) or is just empty.
- Python Algorithm - To solve the sudoku
- The backtracking algorithm to solve a sudoku is very slow, so I wrote another algorithm (check solve.py).
- The algorithm can solve easy, medium and hard sudokus.
- You can also check the code for training the model on colab here: https://colab.research.google.com/drive/1sAMzIrJAwyV1WjHtAjDKP5kt3twjX_IY?usp=sharing