Automatic Recognition of Maya Hieroglyphs in 3D

Automatic Recognition of Maya Hieroglyphs in 3D
Date
22 February 2022, 4.00pm - 5.15pm
Type
Seminar
Venue
Online- via Zoom
Description


In our talk we summarize results of an interdisciplinary cooperation in the field of digital epigraphy between the project "Text Database and Dictionary of Classical Maya" (TWKM) and a team specialized in script research with the eHumanities working group of the Institute for Computer Science at Halle University, which continues the work of the Forensic Computational Geometry Laboratory (FCGL), IWR, Heidelberg University. At the center of the cooperation between epigraphic research and applied computer science is the automatic recognition of characters from the hieroglyphic script of the Classic Maya culture in 3D. This method has been successfully used for several years in the automatic recognition and transliteration of cuneiform characters on clay tablets, which were documented by the same type of high-resolution 3D scanners. In the course of the work for the TWKM has documented Maya inscriptions in 3D and, together with the project partners, tested the possibilities whether the hieroglyphic script of the Maya is suitable for automated recognition. Similar to cuneiform script in 3D the Maya inscriptions feature around 1000 elements. In digital epigraphy the approach is typically based on manual encoding and linking of the texts (e.g. using TEI XML, RDF). The work presented is an approach using machine learning to automatically compute similarities between glyphs resulting in a similarity network. This is achieved using curvature visualizations computed from the 3D measurement data using the Free and Open Source (FOSS) GigaMesh Software Framework. In this sense, an Optical Character Recognition (OCR) for writing in 3D is created. The major goal of the collaboration is the establishment of a highly automated process to create a digital twin of the catalog of Maya script characters to improve the understanding of variants and character morphology.

SpeakersChristian Prager (Bonn) and Hubert Mara (Halle)


This seminar is co-hosted by the Digital Humanities Research Hub, University of London, UK, and Star-UBB Institute of Advanced Studies, University Babeș-Bolyai, Cluj Napoca, Romania.


All welcome, but advanced registration is required.

Contact

SAS Events Office
sas.events@sas.ac.uk
Email only