Work with Deep Learning Studio projects

Available with Image Server

Deep Learning Studio projects contain all the relevant information about the deep learning analysis to be performed. Follow the workflows below to create a project and maintain the status of the project.

Create a Deep Learning Studio project

When beginning a deep learning process with Deep Learning Studio, the first step is to create the project.

The Deep Learning Studio project becomes an item in your organization to maintain a record of the progress and steps that have been performed.

1. Sign in to ArcGIS Enterprise.
2. From the App Launcher, open Deep Learning Studio.

Note:

If the Deep Learning Studio button is not visible on the App Launcher, review the requirements to enable it in ArcGIS Enterprise.

3. Click Create.
4. Type a name for the project in the Project name text box.
5. Choose one of the three project types below:
• Object detection
• Pixel classification
• Object classification
6. Type keywords in the Tag section.
   1. Type the keywords on the dialog box.
   2. Add a comma or press Enter after each tag.
7. Type a short summary of the analysis to be conducted in the Summary section.
8. Click Create to create the Deep Learning Studio project.

Tip:

The Create button is not available until a project name is provided; the other sections are optional.

Edit the project configuration

Once you create a Deep Learning Studio project, you can edit some of the characteristics with additional information.

1. From the App Launcher, open Deep Learning Studio.
2. Select the Deep Learning Studio project to be edited.
3. Click the View/Edit configuration button to open the project configuration.
4. Select the section you want to edit and make the changes.
5. Click Apply.

   At 11.1, Deep Learning Studioallows you to add or remove custom work units after the project has been configured. Prior versions of Deep Learning Studio do not allow the addition or removal of custom work units.

Configure the project for training

When you select the Prepare training data step for the first time, the project must be configured for sample training.

1. From the App Launcher, open Deep Learning Studio.
2. Select the Deep Learning Studio project to begin configuring it for training.
3. On the step choice page, select Prepare training data.

   Tip:

   A prompt to configure the project for preparing training data appears if it has not been configured previously.

4. In the prompt, click Yes.
5. For Choose imagery, select the imagery layer to use as input for the training sample collection.

Available imagery layers are listed in the menu based on the imagery layers that have been created in your organization. Imagery collections in the data store are also available for use as input.

6. Click Next.
7. Create the training sample label schema.
   • Click the Add button to create the first training label.

     Note:

     A child label can be added to the existing label to create a label hierarchy to support your project.

   • If you have existing label schema, click the Import button to import it.
8. Optionally, click the Change Symbology button next to the label to change the symbology of the label.

Tip:

Once the label schema is complete, determine whether the schema will be reused. If it will be reused, consider exporting the label schema to use in other projects.

9. Click Next.
10. Start the Invite members step to select the groups that will be collaborating on the training sample collection process.

Tip:

Expand the groups to see the privileges of the group members in each group.

Using groups is optional in the project and each step can be completed without selecting a group.

Tip:

Groups must be created prior to configuring the project. If no groups are visible, either there are no groups created for your organization or they are not visible to the user account.

11. Click Next.
12. In the Set up work units step, choose the configuration for the work units. Choose the preferred configuration for the project.
You can configure the Grid system and Custom work units options by clicking Configure next to each option.
13. Click Save to complete the project configuration.

Mark the project as completed

Deep Learning Studio projects store all of the steps that have been completed. Once all of the steps are completed, the project owner can mark the project as completed.

1. From the App Launcher, open Deep Learning Studio.
2. For the project that has been completed, click the project update button .
3. Choose Mark as complete.
An icon is added to the project to indicate that it is complete. You can use the completion indicator to filter the projects.