After producing training data and training a neural network, the network can be applied to all images with statistical parameters fitting to the corresponding training data.

Use GeoDict-AI Segment Grayscale Image, to apply neural networks trained to segment 3D-scans (see also Taha and Hanbury, 2015).