Dataset Card for DocLayNet v1.2

Dataset Description

• Homepage: https://developer.ibm.com/exchanges/data/all/doclaynet/
• Repository: https://github.com/DS4SD/DocLayNet
• Paper: https://doi.org/10.1145/3534678.3539043

Dataset Summary

This dataset is an extention of the original DocLayNet dataset which embeds the PDF files of the document images inside a binary column.

DocLayNet provides page-by-page layout segmentation ground-truth using bounding-boxes for 11 distinct class labels on 80863 unique pages from 6 document categories. It provides several unique features compared to related work such as PubLayNet or DocBank:

1. Human Annotation: DocLayNet is hand-annotated by well-trained experts, providing a gold-standard in layout segmentation through human recognition and interpretation of each page layout
2. Large layout variability: DocLayNet includes diverse and complex layouts from a large variety of public sources in Finance, Science, Patents, Tenders, Law texts and Manuals
3. Detailed label set: DocLayNet defines 11 class labels to distinguish layout features in high detail.
4. Redundant annotations: A fraction of the pages in DocLayNet are double- or triple-annotated, allowing to estimate annotation uncertainty and an upper-bound of achievable prediction accuracy with ML models
5. Pre-defined train- test- and validation-sets: DocLayNet provides fixed sets for each to ensure proportional representation of the class-labels and avoid leakage of unique layout styles across the sets.

Dataset Structure

This dataset is structured differently from the other repository ds4sd/DocLayNet, as this one includes the content (PDF cells) of the detections, and abandons the COCO format.

• image: page PIL image.
• bboxes: a list of layout bounding boxes.
• category_id: a list of class ids corresponding to the bounding boxes.
• segmentation: a list of layout segmentation polygons.
• area: Area of the bboxes.
• pdf_cells: a list of lists corresponding to bbox. Each list contains the PDF cells (content) inside the bbox.
• metadata: page and document metadetails.
• pdf: Binary blob with the original PDF image.

Bounding boxes classes / categories:

1: Caption
2: Footnote
3: Formula
4: List-item
5: Page-footer
6: Page-header
7: Picture
8: Section-header
9: Table
10: Text
11: Title

The ["metadata"]["doc_category"] field uses one of the following constants:

* financial_reports,
* scientific_articles,
* laws_and_regulations,
* government_tenders,
* manuals,
* patents

Data Splits

The dataset provides three splits

• train
• val
• test

Dataset Creation

Annotations

Annotation process

The labeling guideline used for training of the annotation experts are available at DocLayNet_Labeling_Guide_Public.pdf.

Who are the annotators?

Annotations are crowdsourced.

Additional Information

Dataset Curators

The dataset is curated by the Deep Search team at IBM Research. You can contact us at deepsearch-core@zurich.ibm.com.

Curators:

• Christoph Auer, @cau-git
• Michele Dolfi, @dolfim-ibm
• Ahmed Nassar, @nassarofficial
• Peter Staar, @PeterStaar-IBM

Licensing Information

License: CDLA-Permissive-1.0

Citation Information

@article{doclaynet2022,
  title = {DocLayNet: A Large Human-Annotated Dataset for Document-Layout Segmentation},
  doi = {10.1145/3534678.353904},
  url = {https://doi.org/10.1145/3534678.3539043},
  author = {Pfitzmann, Birgit and Auer, Christoph and Dolfi, Michele and Nassar, Ahmed S and Staar, Peter W J},
  year = {2022},
  isbn = {9781450393850},
  publisher = {Association for Computing Machinery},
  address = {New York, NY, USA},
  booktitle = {Proceedings of the 28th ACM SIGKDD Conference on Knowledge Discovery and Data Mining},
  pages = {3743–3751},
  numpages = {9},
  location = {Washington DC, USA},
  series = {KDD '22}
}