Augmented_Reality_Sudoku_Solver

Overview

The project consists of 3 parts:

1. Computer Vision - To detect and extract a sudoku from a camera input.
2. Deep Learning - To predict whether each cell contains a number (1 to 9) or is just empty.
3. Python Algorithm - To solve the sudoku

Notes

1. The backtracking algorithm to solve a sudoku is very slow, so I wrote another algorithm (check solve.py).
2. The algorithm can solve easy, medium and hard sudokus.
3. You can also check the code for training the model on colab here: https://colab.research.google.com/drive/1sAMzIrJAwyV1WjHtAjDKP5kt3twjX_IY?usp=sharing