classify pixels using deep learning outputs 32 bit blank raster instead of 8 bit arcgis pro 3.0.2

[Nika]

Hi

I'm trying to extract land parcel boundaries using HED edge detection model. Input raster is 8 bit RGB drone ortho imagery in tiff format. I have exported training data and trained model  successfully, but when trying to classify pixels using deep learning, tool generates 32 bit blank raster. Anyone with similar experience, or any suggestions what may be a problem? 

Thank you in advance!

[ShivaniPathak]

Hi @Nika, the output of Classify Pixels Using Deep Learning tool is always generated as 32 bit depth. In the screenshot I can see you have used a model argument bit_depth = 8, bit depth is not a valid argument and the tool doesn't supports this model argument. Can you try running the tool again without bit_depth model argument?