EduceLab-Scrolls

EduceLab-Scrolls is an open dataset that facilitates the noninvasive reading of Herculaneum scrolls, containing volumetric X-ray CT images of scroll fragments along with 2D image labels for ink detection model training. The tool aims to recover hidden text from ancient papyri, supporting advancements in machine learning and computer vision through competitions like the Vesuvius Challenge. Notable features include comprehensive data from two decades of research and the ability to reveal previously obscured characters within the scrolls.