This is project is an OCR wich could recognize and resolves SUDOKU.
git clone https://github.com/Helyosis/OCR_EPITA.git
cd OCR_EPITA/src
make
git clone https://github.com/Helyosis/OCR_EPITA
cd OCR_EPITA/src/ && make
cd UserInterface/ && make
Commande line:
./main 3.14.15 help:
[ Image mode specific options ]
-i file: Specifiy the input file (required)
-o file: Specify the output file (default: out.bmp)
--show: Show the image being processed, one step at a time
[ Train mode specific options ]
-n nb: Specifiy the number of iterations to train the neural net with (default is 100 000)
-o file: Specify the output file to save the neural network
--batch-size s: Specify the numbers of elements in a minibatch size (default 100)
--nb-images n: Specify the number of image to train with. (default 8228)
--learning-rate n / --step-size n: Specify the step size (default 0.25)
[ General options ]
-v: Increase the verbose level (default 0), can be used up to 3 times
--mode mode: Specify the mode to use. Can be one of IMAGE/TRAIN/GUI (default is GUI)
-h / --help: Show this help and quit
Graphical Interface:
cd OCR_EPITA/src/UserInterface
./interface
Then select your sudoku grid and click on the "solve" button and your grid will be solved automaticaly.
We are making a User Interface, to give a better usage of our tool.