Help prepping sample data for image classifier

Hi Everyone,

I’m trying to train a classifier using data from multiple raster layers that represent 2 different sampling periods of the same landscape. I have 2 shapefiles with polygons representing sample locations of different classes, as well as multiple rasters that correspond to each shapefile. I wanted to make sure I prep/input these files correctly into the gui (I’m using Orfeo in QGIS). Is there a good tutorial available for using multiple data sets like this to train a classifier?



I think the best ressource to get started with classification in OTB is the Cookbook recipe, in particular this part on classification using several image. The recipe uses the command line interface, but the parameters are the same in the QGIS interface.

I am not sure what you are trying to do, you have one shapefile for the first sampling period, and one shapefile for the other ? And N rasters corresponding to shapefile one, and M rasters corresponding to shapefile two ? with N=M (i.e. the rasters depict the same thing at two different periods)?

Or do you have N rasters corresponding to the two sampling period, and two shapefiles with sample location, one for training and one for validation ?