Labeling

Label types

To create or modify label types, click on the button on the top left of the Spectral Explorer page to bring down a dropdown menu. Select Label types to open the label types dialog.

• By default, no label types are created, but in our screenshot below, we have a Default collection. Click the add button to create a new collection, and then click on the + icon to the right of the new collection to create a new label type.

To rename a label or collection, double click its name or right click and select Rename.

Labeling settings

The label settings are near the draw tools and can change the following:

• No class: Labels will by default have no class assigned, but you can select a class from the dropdown to always assign it.
• Exploratory: Labels will initially be exploratory, which means they are only visible in the current workspace. By setting them to Finalized, they can be added to datasets and will be visible in any workspace.

Manual labeling

To manually create labels, click on one of the draw tools at the top of the canvas. As you can see below, we have selected the lasso tool. Zoom in with the mouse wheel and use CTRL + click (or middle mouse button) to pan the image to get a better view of the area you want to label, then draw a selection around your target area.

Once you have made your selection, a new selection or label will appear in the sidebar based on your label settings.

Labeling with PCA

To label with PCA, click on the PCA tab in the bottom toolbar below your image, and then click Generate PCA.

Once the PCA generation is complete, you will see a scatterplot below your image, as shown below. Each point represent a subsample of the spectral signatures in your image. The larger the point on the scatterplot, the more spectral signatures are included. This intended to make it easier for users to select pixels with similar spectral signatures.

Use the scroll wheel to zoom in and out of the scatterplot. You can create a label by clicking the lasso icon in the toolbar near the scatterplot and drawing a selection around your target area.

Once you have made your selection, a new selection or label will appear in the sidebar based on your label settings.

• By clicking the new selection, red dots will appear on the scatterplot indicating the label's scatterplot location. You can use this to further refine your selection, or to identify where a label's cluster appears on the scatterplot.
• In our screenshot below, the image is darkened to better see the PCA selection. You can do this by clicking on the "Darken" filter at the top right of the canvas.