Paper | Code (ink-id + Volume Cartographer)
EduceLab-Scrolls: Verifiable Recovery of Text from Herculaneum Papyri using X-ray CT
Stephen Parsons1,
C. Seth Parker1,
Christy Chapman1,
Mami Hayashida1,
W. Brent Seales1
1University of Kentucky
EduceLab-Scrolls is a comprehensive open dataset representing two decades of research effort on the challenge of noninvasively reading the Herculaneum scrolls.
EduceLab-Scrolls contains a set of volumetric X-ray CT images, of both small fragments and intact, rolled scrolls. The dataset also contains 2D image labels, used in the supervised training of an ink detection model.
Using EduceLab-Scrolls we have revealed, for the first time, hidden characters from the interior layers of Herculaneum papyri.
A subset of EduceLab-Scrolls is now available through Vesuvius Challenge, a machine learning and computer vision competition to build on this work and recover more text from the Herculaneum papyri. For more information, visit scrollprize.org.