Deep-Learning-Based-Sudoku-Solver

The project objective is to train a deep learning model that can solve a Sudoku puzzle from an image. The Project has been divided into two tasks.

Task - 1:

To train a model to read the Sudoku puzzle to know which digits there are and where are their placed withing the Sudoku grid.

Task - 2:

To train a model to actually solve the Sudoku , that is to complete the empty cells in the grid with the correct digits.

Files :

• Task - 1

  • Task - 1.ipynb : The Notebook file related to the Task -1 of the project mentioned above.
  • cnn_fully_trained.pth : The saved pytorch model that is trained on the images to identify the digits.

• Task - 2

  • Data Preparation for Task - 2.ipynb : The Intermediate file to make the given images ready for the next part of the project.
  • TASK - 2 ( Training).ipynb : Notebook File to train deep learning networks on the sudoku grids predicted from the task-1 trained models.