Sudoku Solver

Hi, This is Team NullCrew.

Welcome to Sudoku Solver!

Its all about detecting and directly solving sudoku puzzles in VR using OpenCV and TensorFlow.

What is it doing ?

For a single image process, the algorithm works as following :

Grids Extraction

Firstly, the algorithm has to find where the grids are. The pipeline is:

1. Preprocess image to enhance high frequencies.
2. Find lines using Hough Space.
3. Analyse lines to know which can correspond to a grid.

Digits Identification & Grids Solving

Once grids are extracted, for each grid we do:

• Detect shape which can be digits.
• Use CNN do identify digits.
• Create a numeric grid in a table and solve it.

Grids Reconstruction

• Create a virtual image to fill the initial image.
• Add the 2 images together to create the final result.

How to use?

The best way to use the program is to firstly clone the repo.

git clone https://github.com/ibhanu/sudoku-solver.git

Then, open a terminal & go in the repository

cd /path/to/the/folder/sudoku-solver/

The required libraries are specified in requirements.txt. Run

pip3 install -r requirements.txt

Take a picture of your process target (Sudoku Puzzle) and use the command

python3 sudoku_solver_main.py -i path/to/your/file

Add -s 1 to save the output result. It will be save in the directory images_save. In both cases, the result will be displayed in the process.

If you have your own Keras model to identify digits. For that use the argument -mp -mp path/to/your/model

Where do I find weights for the CNN?

The model is already trained with a dataset of 10k numeric digits with data augmentation. It gives a precision of 99.5%

Authors

• Abhay R Muni
• Ashritha C
• Bhanu Pratap