Do wee need Notebook Server for some of the deep learning function?

490
6
03-17-2019 09:05 AM
HN1
by
New Contributor II

Do we need Notebook Server (or Image Server, perhaps?) to use some of the methods in arcgis.learn module?

Single shot detector seems to work without a Enterprise product, but some methods like 'detect_objects' run on the server side (specified by gis parameter) that deep learning module installed.

0 Kudos
6 Replies
DanPatterson_Retired
MVP Esteemed Contributor

are you referring to this documentation?

arcgis.learn module — arcgis 1.6.0 documentation 

0 Kudos
HN1
by
New Contributor II

Hi Dan,

Sorry for poor information.

Yes I mean the documentation you suggested.

Hopefully, I would like to try these functionality with my laptop, but according to the doc, 'arcgis.learn.detect_objects()' or 'arcgis.learn.classify_pixels' seems to run on the server side only.

0 Kudos
EarlMedina1
Occasional Contributor III

The requirements are basically as described here:

So, you'll need Image Server 10.7 as this operation is only just being introduced (the Python API method references the ClassifyPixelsUsingDeepLearning, etc. tasks)

HN1
by
New Contributor II

Hello Earl,

Thank you for providing resources and make it clarify!

At this moment I do not have my own image server license, so try to find out another way.

Thanks again.

0 Kudos
SandeepKumar1
Esri Contributor

Yes you can also do these workflow using ArcGIS Pro and notebooks, it is basically :-

1. Exporting training data from ArcGIS Pro using the tool 'Export Training data for deep learning'.

2. Importing the training data using arcgis.learn in the notebook and train a model, we can then save the model as a deep learning package.

3. Using the save model in ArcGIS Pro to produce results, tools that can be used 'Detect Objects using deep learning' and 'Classify pixels using deep learning'.

0 Kudos
DanPatterson_Retired
MVP Esteemed Contributor

from the link

Note that the deep learning library needs to be installed separately, in addition to the server’s built in Python 3.x library.
0 Kudos