read_dataset_german_konzilsprotokolle

This dataset arises from the READ project (Horizon 2020). Images were provided and enriched under the lead of Dr. Dirk Alvermann (Universitätsarchiv Greifswald - Germany). All in all this dataset contains 8770 trainscribed textlines of handwritten historical documents from the late 18th century. Besides the images and page-files (containing geometric textline information and transcripts), lists dividing the dataset in train and test data are provided (each list element contains the corresponding image, textregion and textline identifiers and therefore an explicit mapping of a list element to a textline is possible). Furthermore sublists of the train list are given.