Chess-FEN

Predicts Forsyth–Edwards Chess Notation from a chess board image, using an SVC model

Dependencies

Python

I tested it on Python 3.7.10 but it should also work with future versions as well (hopefully)

Modules versions

Python module	Version
numpy	1.19.5
cv2	4.1.2
matplotlib	3.2.2
sklearn	0.22.2.post1
joblib	1.0.1

Usage

$ python3 src/main.py --help
usage: main.py [-h] [--predict PREDICT | --train TRAIN]

Predicts FEN given a chess board image

optional arguments:
  -h, --help         show this help message and exit
  --predict PREDICT  The input image file
  --train TRAIN      Path to a floder of training images

Training data

I used this dataset by Pavel Koryakin which is under CC0 license; Big thanks to Pavel Koryakin.