Abstract
The trained neural network can predict the shape and strain features of a sample without the need for time-consuming iterative steps.