Low dosage SEM image processing for metrology applications

Serial cross sectional SEM images of a multilayer ceramic capacitor (MLCC) were acquired using FIB-SEM. A 3D image was reconstructed from 360 images showing Ni electrode, BaTiO3 dielectric and voids. After segmentation, volume ratio of Ni, BaTiO3, and void regions turned out approximately 29 %, 69 %, and 2 %, respectively. Comparing void ratio of multiple Ni electrodes, there are variations between each layer. #CeramicCapacitor #MLCC #FIB #SEM #FIBSEM #electronmicroscopy #ElectronMicroscope Avai

U-Net: Training Image Segmentation Models in PyTorch - PyImageSearch

Semantic segmentation with OpenCV and deep learning - PyImageSearch