Mystery Auto-Generated Vegetation While Using Deep Learning in ArcGIS Pro
Hello and good day to all
I was going through Deep Learning to locate the Palms in Tonga course, and when I turned off aerial imagery layer I noticed that arcgis populated the basemap with "trees"... many of them are where there are palms in the image, but not all, and I managed to detect most of them with my model, and they definitely weren't created by the deep learning model, but what the heck are they?
I can't select or query them, they seem to just be an auto-populated canopy against the basemap...
By the way, the are only sitting where the imagery layer was sitting(pic 3 shows this). Not above or below. They disappear somewhere around 1:4000. They are more accurate in some cases than my model, could be useful to use them!
Anyone have any ideas?
I'm curious to know how you tweaked the settings to even get a somewhat decent result. I've tried a bunch of different variations and the model is almost useless. I was meticulous with the training samples but the performance is bad. See attached results using the default settings in the tutorial.
