An augmented reality sudoku solver using OpenCV.
pip install -r requirements.txt
usage: main.py [-h] [-f FILE] [-s] [-w] [-d]
arguments:
-h, --help show this help message and exit
-f FILE, --file FILE File path to an image of a sudoku puzzle
-s, --save Save image to specified file's current directory
-w, --webcam Use webcam to solve sudoku puzzle in real time
(EXPERIMENTAL)
-d, --debug Enables debug information output
python main.py --file assets/c6.jpg
python main.py --webcam
- Webcam solver cannot detect when a new puzzle has entered the frame, will try to warp the solution of the first puzzle it sees onto any subsequent puzzles
- OCR predictions are very spotty for hand written digits and stylized fonts
- Cannot solve puzzles that don't have a distinguishable four-point outer border