Groundtruth

This dataset contains the so-called groundtruth in PageXML format of the transcriptions that were used for training text recognition models via the Transkribus platform (see the explanation about text recognition). It contains 515 scans that form a representative cross-section of handwritten resolutions of the States General between 1576 and 1796.

The dataset can be downloaded via the digital library Zenodo. More information about the creation of the dataset is also available there.