MateriAlZ Seminar: Ing. Pascal Thome

Abstract
Understanding the relations between microstructure and properties is one of the main goals in material science. Over the history of modern materials characterization, microscopes became an integral part in almost every research activity, where they significantly contributed to the present state of knowledge. The advance of affordable and more powerful computer hardware already revolutionized the workflow of microstructure characterization in the 1980 when digital microscopes replaced old fashioned analog instruments. In recent years, the capability and reliability of advanced computer vision techniques has drastically improved. Artificial neural networks, as well as machine and deep learning approaches have proven their potential in different fields of science, e.g. in applications for autonomous driving cars. Similar methods can be adapted for quantitative image analysis in material scientific projects where they can be used to automatically characterize different types of microstructures with an accuracy that outperforms classical algorithmic computer vision approaches. These new possibilities once again might change the way microstructures are characterized and interpreted on a fundamental level. Dr. Thome presents examples of his current research activities in the fields of digital microscopy and quantitative characterization of microstructures, including advanced methods of orientation imaging techniques based on electron back scatter diffraction as well as automatic data extraction for optical micrographs by utilizing deep learning approaches. He presents different approaches of how to handle the large amount of raw data and compare the object detection performances of multiple network architectures and various hyper parameters. Finally, Dr. Thome to condense the extracted information by transforming it into an ontology-based data base of geometric representations reflecting all relevant relations between microstructural entities. In this way he effectively creates a digital twin of the microstructure for microscopically characterized samples. This novel approach allows him to perform further automatic data extraction and sets us up for future utilization such as big data mining approaches.

Bio
Dr. Ing. Pascal Thome is the group leader for Microstructure Informatics at the Ruhr University Bochum in Germany. His research group focuses on state of the art computer vision technologies which are adapted and applied for automatic microstructure characterization. Long term objectives include generating microstructure data bases that act as a digital twin reflecting all relevant microstructural properties of the investigated samples. In this way Dr. Thome will contribute to the overall tendency for digitalization and the transformation of the scientific modus operandi of materials science into the data driven world of the twenty first century.