Reassembly of 3D archaeological artifacts

Magali Segal-Stolarsky#, Ayellet Tal and Ilan Shimshoni - Department of Information Systems

 Machine Learning
Deep learning
Neural Networks

Digital Humanities

PhD Grant 2020

There exist thousands of archaeological sites around the world. In each site there were found a vast amount of cultural heritage fragments. And all of those fragments need to be reassembled in order to assist the archaeologists in their work. For Magali Segal Stolarsky, Msc University of Haifa, Prof. Ilan shimshoni, University of Haifa, and Prof. Ayellet Tal, Technion  this means that it is possible to use 3D scans of the fragments to reassemble the objects automatically. Magali’s work shows the importance of using 3D geometric algorithms combined with neural nets in order to automatically reassemble the fragments into a complete object without any additional knowledge. This is done by first extracting geometric data from every fragment, then finding possible pairs of fragments that should be aligned one next to the other. Finally using a confidence scoring method the next fragment to be aligned is iteratively chosen. This process ends with a reassembled object. Why Reassembly of 3D archaeological artifacts? We can reduce the execution time of manually reassembly from days to minutes.