Patrologia Graeca (OCR ground truth)

Ground truth manually produced within the scope of the CGPG project (Calfa GREgORI Patrologia Graeca), led by Jean-Marie Auwers (UCLouvain), that aim to OCRize the remaining non-digital versions of the Patrologia Graeca volumes. This dataset compiles annotations from 2021 to 2022, and has been used for the Programming Historian online lesson "Transcription automatisée de graphies non latines" (link to come).

The dataset contains a set of 100 images from the Patrologia Graeca, with their corresponding pageXML files. Annotation has been performed with the Calfa Vision platform.

Different level of annotations are proposed to overcome two tasks: (task1) detection of text regions in Greek (annotations at the region level only) and (task 2) recognition of ancient polytonic Greek (annotations at the line level).

Task 1:
col_greek: 52
col_lat: 54
footnotes: 27
titles: 9

Task 2:
lines: 2.579

Final model of layout analysis is freely usable on the Calfa Vision platform, by choosing the "Greek printed (Patrologia Graeca)" type of project.

The project is sponsored by the ASBL Byzantion, the Fondation Sedes Sapientiae, the Institut Religions, Spiritualités, Cultures, Sociétés (RSCS, UCLouvain) and the Centre d'études orientales (CIOL, UCLouvain) and by a generous donor who wishes to remain anonymous. Other sponsors have recently expressed their willingness to support the project.