Deep Learning for automated phase segmentation in EBSD maps. A case study in Dual Phase steel microstructures

Electron Backscattering Diffraction (EBSD) provides important information to discriminate phase transformation products in steels. This task is conventionally performed by an expert, who carries a high degree of subjectivity and requires time effort. In this paper, we question if Convolutional Neural Networks (CNNs) are able extract meaningful features from EBSD-based data order automatically classify the present phases within steel microstructure. The selected case study ferrite-martensite discrimination U-Net has been as network architecture work with. Pixel-wise accuracies around ~95% have obtained when inputting raw orientation data, while ~98% reached with orientation-derived parameters such Kernel Average Misorientation (KAM) or pattern quality. Compared other available approaches literature for discrimination, models presented here provided higher shorter times. These promising results open possibility on more complex microstructures.