DSS-Segmentation

Segmentation for DSS manuscripts

DSS Fragments Segmentation pipeline The scripts are located in : https://www.dropbox.com/sh/1dc50g802umalg5/AABZ73a6LvgPJZ0kR-Pex21Sa?dl=0

Crop

1. Run GrabCut->SegmentFragment.py

Configuration:

a. list_to_process – path to text file with all images to process (example files_from_nli_fourth_batch.txt)

b. base_path – local path where all outputs will be saved

c. img_path – path to images from list_to_process

the script will produce 2 outputs for each image that are used in downstream scripts.

1. in base_path/fragments – the cropped fragments
2. in base_path/cords – the coordinates of the crop

Japanese paper removal

Run matlab->script_remove_jp_by_dirs.m

Configuration:

1. base_in_dir – the fragments directory that is produced by SegmentFragment.py
2. base_out_dir – your local directory for images without japaneese paper Note that in some of the images the Japanese paper is not removed.

Boundary creation

Run matlab->script_segment_with_gc.m

Configuration:

1. gc_img_dir – fragments folder either from SegmentFragment.py or base_out_dir after japaneese paper removal.
2. File path of all images to process – in line 28.
3. gc_cords_dir – crop coordinates folder from SegmentFragment.py
4. gc_boundary_dir – output folder for boundaries
5. json_output_dir – output folder for boundaries in json format (for Bronson)