Soxavin/ardb-layout-coco-v1

ARDB Daily Bulletin Layout Detection (COCO, folder format)

⚠️ Please use Soxavin/ardb-layout-coco-v2 instead

This v1 repository is the original folder / COCO export and is no longer maintained. The identical data is published in Parquet format — with a working Dataset Viewer and the standard 🤗 datasets loader — at Soxavin/ardb-layout-coco-v2. New users should load v2:

from datasets import load_dataset
ds = load_dataset("Soxavin/ardb-layout-coco-v2")

A document-layout object-detection dataset built from ARDB (Agricultural and Rural Development Bank of Cambodia) daily market-price bulletins — born-digital Khmer-language PDFs, one price table per page. Every page is annotated with bounding boxes for five layout regions (table, text, section header, page furniture, picture).

Layout on disk

train/  valid/  test/
  ├── <page>.jpg                 # page images, 200 DPI (2000 × 2000 px)
  └── _annotations.coco.json     # COCO annotations for that split

Dataset structure

Split	Pages	Boxes	Source documents
train	73	341	24 bulletins
valid	9	42	3 bulletins
test	9	42	3 bulletins
total	91	425	30 bulletins

Classes

id	name	Boxes	Notes
0	Table	91	The price table — exactly one per page, covering the full table including header row and label columns.
1	Text	30	Body / paragraph text.
2	Section-Header	31	Headings and titles.
3	Page-Furniture	182	Page headers and footers.
4	Picture	91	Logos and stamps.

Boxes use the COCO convention: bbox = [x, y, width, height] in pixels.

Data collection & annotation

1. Source. Publicly published ARDB daily market-price bulletins, rendered page-by-page at 200 DPI.
2. Pre-annotation. Candidate boxes were generated automatically with a document-layout detector, confidence-filtered, and mapped to the five-class set above.
3. Human correction. Every page was reviewed and corrected in Roboflow: fragmented table regions merged to one box per page, edges tightened, mislabels fixed, spurious boxes removed.
4. Splitting. Splits are assigned by document (no bulletin's pages span splits) to prevent near-duplicate-page leakage.

Limitations

• Single template. All bulletins share one recurring page layout, so validation/test measure annotation quality rather than cross-design generalization.
• Small scale. 91 pages / 30 documents.

License

The annotations (bounding boxes, class labels, metadata) are released under CC BY-NC 4.0 — free for non-commercial research with attribution. The page images are reproductions of bulletins published by ARDB, included for research use only; users are responsible for complying with the source documents' terms. This dataset is not affiliated with or endorsed by ARDB.