Segmentation of lath-shaped bainite in multiphase steels

Multiphase steel sample

Multiphase steel sample

  • Samples of a multiphase steel with polygonal ferrite, bainite and carbon-rich second phases
  • Preparation and contrasting with nital etching
  • Images were taken under a light microscope (LM) and scanning electron microscope (SEM)

Task definition:

  • Development of an segmentation routine for lath-shaped bainite

Realization:

  • Use of Deep Learning (semantic segmentation)

  • Including ML models with data sets from light microscope and scanning electron microscope images

  • On the base of correlative microscopy (LM/SEM and electron backscatter diffraction) for objective ground truth

Results:

  • Very good model performance for LM and SEM images (pixel accuracy, intersection over union)

  • Low deviations of the phase components (1-2%)

  • Good segmentation even with new, unseen images

 

DL segmentation of lath-shaped bainite– LiMi

DL segmentation of lath-shaped bainite – SEM

Areas of application:

  • Automated, objective and reproducible recording of the lath-shaped bainite
  • Microstructure analysis as a basis for process-structure-property correlations

Partners for cooperation:

  • Fraunhofer Institut für Werkstoffmechanik, Freiburg
  • Department of Materials Science and Engineering, Carnegie Mellon University, Pittsburgh
  • Aktien-Gesellschaft der Dillinger Hüttenwerke, Dillingen/Saar

References:

  • A.R. Durmaz, M. Müller, B. Lei, A. Thomas, D. Britz, E.A. Holm, C. Eberl, F. Mücklich, P. Gumbsch, A deep learning approach for complex microstructure inference, Nat. Commun. 12 (2021) 1–15. Link to study

Contact

Dr.-Ing. Dominik Britz

Deputy Head MECS Saarbrücken

+49 681 302 70540
d.britz@mec-s.de
LinkedIn

Adrian Thome, M.Sc.

You are interested in working with us?

Feel free to contact us! We look forward to talking to you and finding out how we can help you with your project.