scripts.get_images_from_muscima module is a script that copies out the images for which MUSCIMA++ provides symbol annotations from a download of the CVC-MUSCIMA staff removal dataset.

You have to download this dataset first and provide a path to its root directory (meaning the directory which contains subdirs for the individual CVC-MUSCIMA distortions) to this script. Either supply it directly using the -r option, or set a CVC_MUSCIMA_ROOT environmental variable.

Example invocation: -o ./images -i 4:10 17:8 5:12 21:10 34:3

MUSCIMA++ 0.9 provides a file with the writer:number pairs for its 140 annotated images in this format, which you can feed to the script with: [...] -i `cat path/to/MUSCIMA++/specifications/cvc-muscima-image-list.txt

For an overview of all command-line options, call: -h