- Python3
- inkscape pdf2svg texlive-font-utils
sudo apt install inkscape pdf2svg texlive-font-utils
- Node
- svgo
sudo npm install -g svgo
- multiprocessing-logging numpy opencv-python svgpathtools svgwrite tqdm
python3 -m pip install -r requirements.txt
Convert .eps file to .svg and .png.
convert.py dirs [dirs ...] [-h] [--num-workers NUM_WORKERS] [--log LOG_FILE]
Arguments:
dirs Directories that stores .eps files.
Optional Arguments
-h, --help Show help message and exit.
--num-workers NUM_WORKERS Number of processes. 0 for all available cpu cores. Default to 0.
--log LOG_FILE Path to log file. Default to "convert.log".
Perform segmentation on .svg and .png files.
segmentation.py [-h] [--num-workers NUM_WORKERS] [--log LOG_FILE]
[--conf CONFIDENCE_FILE] [--no-optimize]
[--export-contour] [--export-mask]
dirs [dirs ...]
Arguments:
dirs Directories that stores .svg&.png files.
Optional Arguments
-h, --help show help message and exit
--num-workers NUM_WORKERS Number of processes. 0 for all available cpu cores. Default to 0.
--log LOG_FILE Path to log file. Default to "segmentation.log".
--conf CONFIDENCE_FILE Path to segmentation confidence file. Default to "seg_conf.json".
--no-optimize Dont't use svgo optimization. This will produce larger svg files but cost much less time. Default to use optimization.
--export-contour Export contour segmentation results. Default to not export.
--export-mask Export morphed mask for debug use. Default to not export.
python3 convert.py dir1 dir2 ---num-workers=8 --log test_convert.log
python3 segmentation.py dir1 dir2 --num-workers=8 --log test_seg.log --conf test_conf.json
segmentation_bbox.py provides API segmentation_bbox for cropping a bounding box area from an SVG file. An example usage can be found in __main__ function.