Select to view content in your preferred language

DL classification of multispectral images in AGP

05-14-2020 10:45 PM
New Contributor

Hi everyone, I am trying to train a UNET pixel classification (instance segmentation) model using chips extracted from a multispectral image (7 bands). I am using the ArcGIS Pro Train Deep Learning Model Tool to do that. The training succeeds only if the chips are extracted from a 3-band version of the image. Once I use the 7-band chips, it fails. I don't mind reducing the number of bands. However, I wanted to know if multispectral images (more than 3 bands) are supported natively in the ArcGIS Pro implementation? Any suggestions to overcome this issue?  Thanks in advance for your assistance

0 Kudos
0 Replies