The Deep Learning Guide Map helps you through the deep learning process. It highlights the tools you may want to use at each step, and provides quick access to examples, system compatibility testing, and TensorBoard metrics.

To begin, select Deep Learning > Deep Learning Guide Map in the ENVI Toolbox.

Click the various buttons based on what you want to do. Use the Back and Forward buttons to navigate to different steps in the process. The Guide Map keeps track of all of the input and output files that you create throughout the process so that you do not have to manually specify them in each step.

In the first panel of the Guide Map, click the button for the type of model you want to create. The options are as follows, with links that take you to the associated topics for details:

• Pixel Segmentation Model: Pixel Segmentation Overview, Train TensorFlow Pixel Models

• Object Detection Model: Object Detection Overview, Train TensorFlow Object Models

• Grid Model: Grid Overview, Train TensorFlow Grid Models

The menu bar of the Guide Map provides additional capabilities, which are described below.

Tools

• Edit TensorFlow model metadata

• Train deep learning models using the ENVI Modeler

• Randomize training parameters using the ENVI Modeler

• Test the Deep Learning installation and configuration

Tasks

Access individual tasks whenever you need them, regardless of where you are in the Guide Map.

Pixel Segmentation

Object Detection

Grid

Training Metrics

View and manage training metrics.