HTR Winter School Vienna 2025 - Carolingian Minuscule Model Improvement

This dataset contains ground truth for 258 pages of Latin manuscripts written in Carolingian Minuscule (9th-12th centuries). It serves as an extension to the Carolingian Minuscule Model (CMM) in Transkribus. It was created by participants of the Carolingian Latin group during the Winter School: HTR of Historical Sources, held in Vienna in 2025.For details, see the official Github Community: https://github.com/HTR-School-Vienna/2025--Carolingian_Latin- Origin of the data Jena, Ms. Bos. q. 19, fol. 179v-184r, 185r-190r, 191r Karlsruhe, BLB, Cod. Karlsruhe 504, fol. 187r-194r Köln, Dombibliothek, Cod. 85, fol. 105r-109r, 114v-116v Leiden, Universitaire Bibliotheken, BPL 141, fol. 7v-9v, 10v-12v, 13v-16r Paris, BnF, lat. 1979, fol. 1r-4v, 12v-18v Paris, BnF, lat. 10911, fol. 3v-13r St. Gallen, Stiftsbibliothek, Cod. 167, p. 10-31 St. Gallen, Stiftsbibliothek, Cod. 561, p. 95-115 Vatikan, BAV, Pal. lat. 261, fol. 105r-115r Vatikan, BAV, Reg. lat. 484, fol. 6v, 8r-15r Wien, ÖNB, Cod. 89, fol. 2v-5r Wien, ÖNB, Cod. 189, fol. 82v-89v Wien, ÖNB, Cod. 427, fol. 75v-82v Wien, ÖNB, Cod. 596, fol. 3r-7r, 9r-9v, 14v-16v Wien, ÖNB, lat. 652, fol. 34r-44r These folios were transcribed by Transkribus using both the private model CMM 2025 v2 (CER 4.77) and the public model Carolingian Minuscule Model CMM 9th-11th c. (CER 5.2). Transcriptions were then manually corrected by the group members.