To work with the deep learning tools in ArcGIS Pro, you need to install supported deep learning frameworks.
To install deep learning packages in ArcGIS Pro, first ensure that ArcGIS Pro is installed. Create a new Python deep learning environment by cloning the default Python environment arcgispro-py3 (while you can use any unique name for your cloned environment, the steps below use deeplearning). Then set up the Python environment provided with your ArcGIS Pro setup. The default Python environment is found in C:\Program Files\ArcGIS\Pro\bin\Python\.
Next, install the following Python packages: Tensorflow, fast.ai, Keras, Pytorch, Scikit-image, Pillow, and Libtiff. Keep in mind you will be installing specific versions of these packages indicated in the installation steps below. The tools only work with these specific versions. Once the packages are installed, you can swap the ArcGIS Pro environment and use the tools.
The deep learning packages cannot be installed using the Python Package Manager. You need to install the packages through the command prompt.
Follow these steps to set up your machine to use the deep learning tools:
- Open the Python Command Prompt window.
- Type the following commands:
- conda create --name deeplearning --clone arcgispro-py3
- activate deeplearning
- conda install tensorflow-gpu=1.14.0
- conda install keras-gpu=2.2.4
- conda install scikit-image=0.15.0
- conda install Pillow=6.1.0
- conda install fastai=1.0.54
- conda install pytorch=1.1.0
- conda install libtiff=4.0.10 --no-deps
- proswap deeplearning
You can now use the deep learning tools.