EBC1 — Environmental Barrier Coating

← Back to Sample Data

Part of Sample_data/Train_test_images/ — JuSPICE’s annotated training and test datasets, each split into train/, val/, and test/ partitions with corresponding pixel-level masks.

Cross-sectional SEM images of a silicon carbide composite (SiC/SiC) environmental barrier coating (EBC) with a thermally grown oxide layer, split into train/, val/, and test/ partitions with corresponding pixel-level binary masks (_annot/ directories). The segmentation task is binary: oxide layer versus background. This dataset was introduced in the NASA MicroNet study and is used here to train and evaluate the NASA MicroNet segmentation model and as the source for the physics-based (PB) synthetic-data examples.

Modality: Scanning Electron Microscopy (SEM) with pixel annotations

Used by: example_nasa_micronet, example_PB

References:

Stuckner, J., Harder, B. & Smith, T.M. (2022). Microstructure segmentation with deep learning encoders pre-trained on a large microscopy dataset. npj Computational Materials, 8, 200. doi:10.1038/s41524-022-00878-5

NASA Pretrained Microscopy Models

See also

• TiO₂ EM Dataset — another annotated SEM segmentation dataset

• SiO₂ Lithiation / Bulk Water — the two video-sequence datasets

• Sample Data — every sample dataset