Convert Marmot Math Dataset to TFD-ICDAR 2019 compatible dataset
The Marmot Math Dataset is a part of Marmot Dataset for mathematical formula identification. Convertation this to TFD-ICDAR 2019 compatible dataset allow to use Evaluation and Visualization tools for a Dataset for Typeset Math Formula Detection.
Usage
marmot2ICDAR.py [-h] --xml_dir 'ground truth' --img_dir image --dest_img_dir DEST_IMG_DIR --dest_math_dir DEST_MATH_DIR
where:
- 'ground truth' - the directory with bounding boxes info .xml files
- image - the directory with images of doc pages
- DEST_IMG_DIR - the directory of directories with ICDAR-style doc image files
- DEST_MATH_DIR - the directory of ICDAR-style math groung truth .csv files (such as TFD-ICDAR2019/TFD-ICDAR2019v2/Train/math_gt/ , for example)