Define Your Labels
Plainsight's annotation tool supports several types of labels to choose from. Choosing a label type depends on the model being used and the intended outcome of the model.
If you plan to use the dataset to train your model in Plainsight, keep in mind that certain label types are not currently supported for model training: Feature Points, Text, Classification, Multi Classification. If you wish to use these label types, you can export your annotations in Plainsight format and train your model outside of Plainsight.

Supported Label Types

    Rectangles: Object detection
    Polygons: Semantic and instance segmentation
    Feature Points: Key point detection for a group (not supported for SmartML model training)
    Text: image captioning (not supported for SmartML model training)
    Classification: Image classification (not supported for SmartML model training)
    Multi Classification: multi-label image classification (not supported for SmartML model training)

Creating a New Label Definition:

    1.
    Navigate to "Label Definitions" tab within a project
    2.
    Enter the name of the new label
    3.
    Select the type of new label
      Finish setup for label type selected (Setup will vary on label type)
    4.
    Click "Save" button

Adding a Sublabel

Rectangles, Polygons, and Feature Point labels have the option of adding sublabels into the label definitions.
Sublabels cannot be used to train models with SmartML
This is useful for having classes associated with higher level instances. Example: drawing rectangles around birds for object detection, but having a class for types of bird within the rectangle (Duck, Geese, etc...).
    1.
    Click "Add Sublabel"
    2.
    Type a sublabel name
    3.
    Choose sublabel type
    4.
    Fill out label options
    5.
    Click "Save"
Supported Sub-label types: Classification, Multi Classification, and Text.

Edit Label

    1.
    Click on the edit icon
    2.
    Make desired changes to the label options
    3.
    Click "Save"

Delete Label Definition

    1.
    Click on the trash-bin icon
    2.
    Click "Yes" or "Confirm" to confirm deletion
Last modified 2d ago