This is the repository for the class EEL7513 - Machine Learning project.
You need to build your own dataset running build_image_dataset script. After running the script
a file chess_dataset.h5 will be on data folder.
The HDF5 file created after run build_image_dataset script has the following structure:
chess_imgs_train: samples dataset to train, with shape (n, 227, 227, 1) wherenis the number of samples.chess_labels_train: chess_imgs_train labels with shape (n, 1) wherenis the number of samples.chess_imgs_test: samples dataset with shape (n, 227, 227, 1) wherenis the number of samples.chess_labels_test: chess_imgs_test labels with shape (n, 1) wherenis the number of samples.
Note: Images converted to grayscale.
chess_labels example
If chess_imgs_train[n] is a black king piece, so chess_labels_train[n] == [1], as well chess_imgs_test and chess_labels_test.
labels = [
'bb',
'bk',
'bn',
'bp',
'bq',
'br',
'empty',
'wb',
'wk',
'wn',
'wp',
'wq',
'wr'
]