Sudoku Computer Vision Solver

A deep neural network trained from scratch. > 99% accurate in various lighting conditions.

Setup

1. Clone this repository (fork it first if you want to deploy to Heroku):

   git clone https://github.com/chrisluedtke/sudoku-cv.git
   

2. Within project repository, create a virtual environment:

   python -m venv env
   

3. Activate the environment:

   Windows:

   env\Scripts\activate
   

4. pip install --user --upgrade pip

5. Install requirements:

   pip install -r requirements-opencv.txt
   

   or

   pip install -r requirements-pi.txt
   

6. Create a .env with contents:

   FLASK_APP=app:APP
   FLASK_ENV=development
   CAMERA=opencv
   

7. Run the web app (the database will be initialized automatically):

   flask run
   

8. Navigate to the locally served page, typically http://localhost:5000/

9. Optional: Create an ipython kernel to use Jupyter Notebook with this environment (see documentation).

   ipython kernel install --user --name=sudoku
   

   • You may also need to copy the .dll files from envLib/site-packages/pywin32_system32 to env/Lib/site-packages/win32