AttributeError: module 'name' has no attribute 'install_deep_learning_model'

[YanLi3]

Hey everyone,

I'm trying to do some deep learning tasks, and I got this error message when I was trying to install the model which was successfully loaded from my AGOL package layer. Anyone knows what's the cause of this and how to fix it out?

The version of my arcgis is 1.8.1, and ArcGIS Pro is 2.6.2 without Image Analyst license. Is this relative to the error? cause it seems happen at tools.py at 7587 line, _tbx.install_deep_learning_model().

Thanks in advance.

Yan

[DanPatterson]

Did you install the framework for you particular version of Pro?

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

[YanLi3]

Hi Dan,

Thank you.

I can do training like ssd.fit, FasterRCNN.fit,... so I think I already install the package of .learn. Anyway I'll read the link and give it a try. Will be back soon.

Yan

[Tim_McGinnes]

Hi Yan,

The install_model function in ArcGIS.learn is used to deploy a deep learning model to an ArcGIS Enterprise Image Server. Is that what you are trying to do?

What sort of model are you using? If you don’t have the Image Analyst extension then some of the object detection models can run the model using the arcgis.learn predict function.

[YanLi3]

Hi Tim,

Thank you.

No I'm not going to deploy it. Plus, it's not necessary to install after calling Model(mymodel) to do object detection. I checked the model info it was good. 

Yes I found I can like run learn.ssd.fit() even without the Image Analyst. 

Now my problem is, I can't run detect_objects(). It seems like the input raster must be a layer from online item. atcpy.Raster output is not accept. While I have no privilege to create an imagery layer to my AGOL, or share from Pro. Stuck here.

[Tim_McGinnes]

The detect_objects function is specifically only for Image Server, which is why it needs a raster layer. And it currently only works in ArcGIS Enterprise and not AGOL.

There is a function within the SSD learn section called predict - it should take a local raster file and return a list of bounding boxes of the detected objects.

[YanLi3]

Thank you Tim. I'll give it a try. Appreciate it so much.