[Virtual Talk] John Hessler on Markov Random Fields, Ancient Maya Ceramics, and 3D Images
Video Ergo Scio: Using Markov Random Fields to Reconstruct Ancient Maya Ceramics and Inscriptions
This virtual seminar will introduce participants to the theory of Markov random fields applied to the reconstruction of ancient Maya ceramics found in archaeological contexts and to the understanding of damaged Maya inscriptions.
The Theory of Markov random fields (MRF) has recently emerged in artificial intelligence research, both as a tool for modeling computer vision, and as a means for making deep and powerful statistical inferences about 3D digital images. These inferences allow for the reconstruction of the underlying objects and scene structure, as well as solutions to such problems as image reconstruction, image segmentation, and the rebuilding of missing parts of damaged 3D archaeological objects. I will discuss my recent work on the reconstruction of ancient Maya pottery from fragments, the complexity of the jigsaw puzzle in computer vision, and the application of newly developed MRF algorithms to the reconstruction of Maya hieroglyphic inscriptions.
John Hessler is the Curator of the Jay I. Kislak Collections of the Archaeology and History of the Early Americas at the Library of Congress. On the faculty of the Rare Book School of the University of Virginia, he teaches the history and structure of the Masoamerican Codex. Interested in the intersection of computation and cultural heritage preservation, he has also taught seminars in computer vision & computational morphometrics at the Kunsthistorisches Institut in Florence, Italy. He is the founder of the BIO-COMP Lab (BCL), where they apply advanced mathematics and computation to challenging problems in archaeology and cultural heritage preservation.
This event is co-sponsored by the Scholars' Lab, the Interdisciplinary Archaeology Program, and the Institute for Advanced Technology in the Humanities (IATH). It is free and open to all, but advance registration is required. Please visit https://cal.lib.virginia.edu/calendar/events/JHessler2021 to register.