DLPK Error 002667 but Deep Learning Frameworks are installed

467
4
11-16-2023 12:54 AM
afyute
by
New Contributor

Hello,

I was trying to run the Detect Object using Deep Learning tool using a deep learning package, but the tool cannot read the .dlpk file. I am getting the "Error 002667: Unable to intialize python raster with scalar arguments". Upon research, apparently it means that I need to have the right Deep Learning Python API (CNTK, TensorFlow, and so on) installed.

However, it's already installed on my ArcGIS. I'm using the Pro Version 3.1 and my Deep Learning Libraries are also 3.1, meaning that I'm using the correct version. I've already tried uninstalling/ reinstalling my Pro and the Deep Learning libraries but the problem persists.

Does anyone have any idea how to fix this?

1.png2.png

0 Kudos
4 Replies
DanPatterson
MVP Esteemed Contributor

Did you select the first two parameters in the dialog prior to selecting the model definition?

Also move your definition file to some other folder than 'downloads'

Also, follow the instructions on their Github site for your version

Esri/deep-learning-frameworks: Installation support for Deep Learning Frameworks for the ArcGIS Syst...


... sort of retired...
0 Kudos
afyute
by
New Contributor

I've tried selecting the first two parameters in the dialog prior to selecting the model definition and moved the definition file from 'downloads' but I still get the Error 002667.

As for the installation of the deep learning frameworks, for Pro version 3.1 you only need to use the correct installer, which I did, but it still doesn't work.

0 Kudos
SusanUSC
New Contributor II

Hi Afyute,

Was this problem solved? I run into the same error after I upgraded Pro to 3.2.2 and deep-learning-essentials 3.2. The model worked in older version. Appreciate any comments. 

0 Kudos
DanPatterson
MVP Esteemed Contributor

002667: Cannot load python module.—ArcGIS Pro | Documentation

it can't find the required module(s) so there is something wrong with your installation or you have remnants of prior installs.

delete your cloned environment and try again in a fresh install


... sort of retired...
0 Kudos