SEM Layer decetion

Use Case:

Automatically detect SEM cross section layer for plotting or further SEM image processing.

Model:

Use YOLOv4, train the model with SEM cross-section images with layers. Pre-trained weights is saved as yolov4-custom_best.weights in model folder. It's trained on only one class (Layer), but can be generalized for multiple useages since the weights is obtained by training the COCO dataset.

Example:

original SEM cross section image:

SEM cross section image with layer detection:

Recreation notebook.

To use the Layer Detection, see Example.ipynb, which can be opened using Colab.

Example: https://github.com/tkm22/SEM-Layer-Detection/blob/master/Example.ipynb