Hi there,
Im using ArcGIS Pro 2.9 and just curious about how exactly training data chips are actually created from labeled objects?
For example, if working with labeled polygons and selecting a tile size 256x256, are chips created by randomly selecting a pixels within the polygons and drawing 256 pixels box around them? If so, how are the total number of chips calculated? If using a stride half the size of the tile, should there always be more (2x?) chips than labeled polygons?
cheers,
Javier