This package is intended to provide an easy way to store Labelings for image segmentation. The segmentation information is stored in a BSON file for fast I/O on demand, but can also be exported to JSON for human readability.
-
Create a Labeling object with one of the static methods or just init normally.
-
Add one or more images(iterate_over_images(), add_image(), or even patches of an image (add_segments()).
Note: If you decide to add patches, make sure that no segment in a patch is overlapping the edge of the patch. Otherwise, the other parts of the segment will be considered an unique entity.
-
(Optional) Add any kind of metadata you want to the Labeling as additional information supplied by you
-
Get the Labeling for further use(get_result()) or save (save_result()) it to a specified path. Both methods return an image and a bson container.
The project is open-source. If you find problems or have ideas for improvement, please create an issue on GitHub, or contribute yourself!