KIT | KIT-Bibliothek | Impressum | Datenschutz

A deep learning approach for complex microstructure inference

Durmaz, A. R. ORCID iD icon; Müller, M.; Lei, B.; Thomas, A.; Britz, D.; Holm, E. A.; Eberl, C.; Mücklich, F.; Gumbsch, P.

Abstract:

Automated, reliable, and objective microstructure inference from micrographs is essential for a comprehensive understanding of process-microstructure-property relations and tailored materials development. However, such inference, with the increasing complexity of microstructures, requires advanced segmentation methodologies. While deep learning offers new opportunities, an intuition about the required data quality/quantity and a methodological guideline for microstructure quantification is still missing. This, along with deep learning’s seemingly intransparent decision-making process, hampers its breakthrough in this field. We apply a multidisciplinary deep learning approach, devoting equal attention to specimen preparation and imaging, and train distinct U-Net architectures with 30–50 micrographs of different imaging modalities and electron backscatter diffraction-informed annotations. On the challenging task of lath-bainite segmentation in complex-phase steel, we achieve accuracies of 90% rivaling expert segmentations. Further, we discuss the impact of image context, pre-training with domain-extrinsic data, and data augmentation. Network visualization techniques demonstrate plausible model decisions based on grain boundary morphology.


Verlagsausgabe §
DOI: 10.5445/IR/1000140450
Veröffentlicht am 06.12.2021
Originalveröffentlichung
DOI: 10.1038/s41467-021-26565-5
Scopus
Zitationen: 12
Web of Science
Zitationen: 10
Dimensions
Zitationen: 20
Cover der Publikation
Zugehörige Institution(en) am KIT Institut für Angewandte Materialien – Computational Materials Science (IAM-CMS)
Publikationstyp Zeitschriftenaufsatz
Publikationsjahr 2021
Sprache Englisch
Identifikator ISSN: 2041-1723
KITopen-ID: 1000140450
Erschienen in Nature Communications
Verlag Nature Research
Band 12
Heft 1
Seiten Art.Nr. 6272
Vorab online veröffentlicht am 01.11.2021
Nachgewiesen in Web of Science
Dimensions
Scopus
KIT – Die Forschungsuniversität in der Helmholtz-Gemeinschaft
KITopen Landing Page