cuts off lines or words of a book from the google 100 books project
then recognizes, alignes and trains these lines or words

usage:
    train-<lines|words>.sh <google-book-dir> <model-for-alignment> <output-dir-for-images> <new-model>

