Datasets:
license: cc-by-4.0
language:
- cop
task_categories:
- image-to-text
tags:
- coptic
- handwritten-text-recognition
- HTR
- historical-documents
- line-segmentation
pretty_name: SCAM – Sahidic Coptic Ancient Manuscripts
SCAM – Sahidic Coptic Ancient Manuscripts
Project page: scam.silviacascianelli.com
SCAM is a dataset of handwritten Sahidic Coptic manuscript pages, with manually annotated text-line polygons and HTR transcriptions. It is split into two subsets:
| Subset | Source | Pages | Lines |
|---|---|---|---|
| SCAM-A | Biblioteca Apostolica Vaticana | 24 | 1,445 |
| SCAM-B | Biblioteca Nazionale di Napoli | 15 | 1,795 |
| Full | Combined | 39 | 3,240 |
Each split (train / eval / test) is encoded in the annotation XML files.
Repository contents
annotations/
page_annotations_RO.xml # SCAM-A line polygons + transcriptions (CVAT 1.1 format)
page_annotations_NA.xml # SCAM-B line polygons + transcriptions (CVAT 1.1 format)
transcriptions/
scam_a/ *.rtf # Page-level transcriptions for SCAM-A
scam_b/ *.rtf # Page-level transcriptions for SCAM-B
The XML annotation files are in CVAT Image 1.1 format.
Each <image> element contains <polygon> elements, one per text line, with:
points– polygon vertices (x,y coordinates) within the page image<attribute name="Content">– the HTR transcription of that line
⚠️ Images not included
Due to intellectual property restrictions, the original manuscript page images are not distributed in this repository. They can be obtained as described below.
SCAM-A — Biblioteca Apostolica Vaticana
Images can be found in the Vatican Apostolic Library Digital Repository:
🔗 https://digi.vatlib.it/mss/Borg.copt
Please inquire with the library regarding the terms under which the images may be used.
The specific fascicles used in SCAM-A are:
MSS_Borg.copt.109.cass.XXV.fasc.120MSS_Borg.copt.109.cass.XXIX.fasc.159
The image filenames in the annotations map directly to the folio identifiers used in that collection (e.g. Borg.copt.109.cass.XXIX.fasc.159_0003_fa_0001r).
SCAM-B — Biblioteca Nazionale di Napoli
Images from the National Library of Naples must be requested directly from the library:
🔗 https://www.bnnonline.it/it/1/home
Further support
For questions about the dataset, annotation methodology, or to request additional information, please contact the authors — their email addresses are provided in the accompanying paper.