In museums, collections, libraries and archaeological excavations the daily use of 3D scanners is rapidly increasing to acquire precise 3D models of tangible heritage. This documentation in 3D is the basis of in-depth analysis of objects like cuneiform tablets, roman ceramics, Greek sealings or medieval inscriptions. The visualization and extraction of characters, symbols, traces of wear or usage is key to provide information about e.g ancient trade networks or manufacturing techniques. Therefore we will introduce the core concepts of 3D data processing to machine learning for character recognition on the example of cuneiform script. A second example are similarity metrics for Minoan and Mycenaean sealings. Especially the visualization of a similarity network of Maya glyph s will illustrate its assistance in the Text Database and Dictionary of Classic Mayan (TWKM, housed at the University of Bonn). Additionally we will have an analytic hermetic discussion of the underlying computational methods.
Sample 3D-data for the hands-on session will be provided, which will be inspected and visualized with our GigaMesh Software Framework. Linux packages and Windows binaries are freely available at https://gigamesh.eu and can be installed in advance. The sample data will be prepared to be suitable for processing on standard PCs/Laptops. Comprehensive large 3D-datasets for cuneiform script are available in advance via the Heidelberg University Library platforms HeiDATA and HeidICON. Please note, that these high-resolution real world data-sets require a certain minimum of CPU and GPU computing power.