A new approach to deciphering cuneiform tablets using the OCR(Optical Character Recognition) method has been developed by a team at Martin Luther University Halle-Wittenberg (MLU). This work draws on methods and technologies developed by Timo Homburg at i3mainz as part of the collaborative project “Digital Edition of the Cuneiform Texts from Haft Tappeh,” a joint initiative between Mainz University of Applied Sciences and Johannes Gutenberg University Mainz. Homburg presented the method at the 21st Eurographics Workshop on Graphics and Cultural Heritage (GCH 2023), which took place September 6–8 in Lecce, southern Italy. The team was awarded the Best Paper Award for this paper. We extend our warmest congratulations.
Cuneiform tablet, Photo: University of Halle / Maike Glöckner, All rights reserved
The artificial intelligence uses 3D models of cuneiform tablets from the Hilprecht Collection, which were prepared by Timo Homburg. Based primarily on these 3D models, Ernst Stötzner developed an artificial intelligence system at Martin Luther University Halle-Wittenberg (MLU) as part of his master’s thesis. He was supervised by Hubert Mara. The AI uses the OCR method, in which software identifies characters and text in images. This is typically based on photographs or 2D scans. For the cuneiform tablets, the team hoped to achieve better results using 3D models, since light and viewing angles affect how well the characters can be recognized. In fact, the AI was able to reliably recognize the characters of two languages.
Scan of a tablet, Photo: University of Halle / Maike Glöckner, All rights reserved
In addition to GCH 2023, the team also presented its results at the International Conference on Computer Vision( ICCV23), the most important international event on computer vision, which took place in Paris from October 2 to 6, 2023.
