Sudoku OCR

Usage

GUI

The GUI can be used by executing the executable with the follwing name:

$ ./sudoku-ocr-gui

CLI

The CLI automatically performs the image processing and grid splitting in verbose mode. Here is how to use it:

$ ./sudoku-ocr --help
Usage: ./main file [options]

Options:
        -o <file>     Save the output into <file>. (PNG format)
        -r <angle>    Rotate the image according to the specified <angle>.
        -m <file>     Save the mask image into <file>.  (PNG format)
        -v <path>     Verbose mode: save every step of the process in the folder <path>.

You can then use the network to process each digit individually

$ ./network test <path>

Network

You can generate a training set by executing the script in the following path:

Note: Please make sure you've got python3 installed along with Pillow (PIL)

$ cd ./neural_network/training_set_generation

$ python3 generate.py

Then, you can train it using

$ ./network train

You can then use the network to process each digit individually

$ ./network test <path>

Build

• Make sure you've installed make, gcc, sdl2, sdl2_image and gtk3

• Clone the repo

• Execute make at the root of the repo

Contents of the Repository

Project Reports

• ./rapport_1.pdf : First Defense Report
• ./rapport_2.pdf : Second Defense Report

Pictures

• ./assets/grids/image_0[1-6].jpeg : Pictures Used for Testing