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).

To start, load the image to analyze by selecting File → Load Volume File … and browsing for the *.grw image to improve. Then, select Segment Grayscale Image from the pull-down menu in the GeoDict-AI section.

Click the Options’ Edit... button.

The Segment Grayscale Image dialog is organized in two tabs: AI Options and GPU Options.