Cropped Cuneiform Sign Collection: A Comparative Dataset of Neo-Babylonian and Neo-Assyrian Scripts

This dataset presents the most common 25 cropped signs (a, an, diš, gar, ud, mu, na, ki, ma, u₂, i, meš, šu₂, nu, aš, ni, igi, šu, e, ta, ka, ti, ru, bad, giš) in the Neo-Babylonian and Neo-Assyrian eras. The dataset results from the union of three open-source datasets, as explained in detail below.A main contribution of this work is the publicly available eBL dataset. 16,943 cropped images from the eBL dataset are used here.The Late Babylonian Signs (LaBaSi) project, focuses on signs from the Neo-Babylonian period. 3,968 cropped images from the LaBaSi dataset are used in this dataset.Cuneiform Digital Palaeography Project (CDDP) dataset. The CDPP dataset consists of a series of snippets of images and an Excel file with metadata. With the help of the eBL database, we were able to extract the class for the CDPP signs by querying the museum number against the eBL’s Fragmentarium. 1,441 cropped images from the CDPP dataset are used in this study.