Jan Philipp Bullenkamp

PhD Project: Semantic Segmentation of 3D Scanned Cultural Heritage Artifacts

This project develops automated methods to segment 3D scans of cultural heritage objects, focusing on lithic artifacts and cuneiform tablets. For lithics, discrete Morse theory segments scars to aid in understanding tool production. For cuneiform tablets, a watershed-based method can detect wedges. Further, a mesh-based neural network is trained on synthetic and experimentally generated data to perform wedge segmentation.

Florian Linsel

PhD Project: Graph Modeling of Segmented Surface Features for Investigating Operational Sequences and Lithic Technologies in 3D

This PhD project focuses on a multifaceted approach to derive graph models of surface features (e.g. scars, ridges and cortex) from manually, semi- and fully automatically segmented 3D models. These resulting graph models are then used to predict the temporal relationships between surface features for reconstructing operational sequences. Additionally, the graph models can be used to train neural networks to distinguish between lithic technologies using automatically segmented and temporally enriched graph models of experimental knapping sequences.