Datasets:
Improve dataset card
#1
by davanstrien HF Staff - opened
Rewrites the card body only; YAML frontmatter is unchanged byte-for-byte.
Sources used:
- Clerice, T. (2022), 'You Actually Look Twice At it (YALTAi)', Journal of Data Mining and Digital Humanities. arXiv:2207.11230 (Tables 1-4 for source composition, per-class instance counts, split construction and reported scores)
- Zenodo deposits 10.5281/zenodo.6827706 (tabular) and 10.5281/zenodo.6814770 (SegmOnto MSS/EPB)
- SegmOnto guidelines, https://segmonto.github.io/gd/gdZ/ (zone definitions)
- Gabay, Camps, Pinche & Jahan, 'Segmonto', ICDAR 2021
- The repo's own loading script and dataset_infos.json (split image counts, config schemas, archive layout); load failure reproduced against datasets 5.0.0
Removed the table of contents and the empty [More Information Needed] stubs, added a SegmOnto gloss, cross-links between the two YALTAi datasets, the known annotation caveats, and a working route to the data now that script-based loading is unsupported.
davanstrien changed pull request status to merged