OCR-D-LAYOUT-GT-PRImA

A ground truth (GT) dataset created within the DFG project “Robust and high-performance methods for layout analysis in OCR-D” and consisting of 464 pages extracted from historical documents. The data publication comprises a selection drawn from eight datasets published by the PRImA Research Group, which have been deployed in various layout competitions over the past few years (see also https://www.primaresearch.org/datasets). In particular, this data publication compiles a total of 464 page images that are free to reuse from the datasets HBR2013 (105 pages), HDLAC2011 (102 pages), HNLA2013 (57 pages), PRImA2009 (8 pages), RASM2019 (10 pages), RDCL2019 (91 pages), REID2017 (54 pages) and REID2019 (37 pages) together with the corresponding ground truth files that were enriched during the project.