What is the Annotation Toolbar?

The annotation toolbar contains the functions that facilitate data annotation. This includes our extensive annotation tool suite to create pixel-perfect annotations and easily manage them.

Annotation

The Annotation section contains manual buttons for each of the Annotation Tools including intelligent tools like IntelliBrush.

It also contains buttons for editing and deleting selected annotations on the Annotation Canvas. More detail on those functions can be found on Automatic Tools and Manual Tools.

Tags

At the bottom of the Annotator menu is the Tags submenu. This section lists the different tag classes that exist in the dataset. You can click on the different tags before making annotations to change the class that the annotation you are going to make will belong to. By clicking the eye on the right of certain tags, you can remove the display of certain classes' annotations on the image if they exist on the image.

Clicking on the upper right of the box brings you to the tag name altering box. Here, you can add the tag name to the list of tags at the upper half, as well as click on current tags to rename them or remove them with the x next to the tag name.

Objects

The Objects tab lists all the annotated objects that are named by their tag class and which numbered instance it is. The eye on the right allows you to toggle vision of the annotated instance, and the symbol with the two overlapping squares allows you to duplicate the annotation.

Background Class

Assets with no objects or regions of interest can be marked with a Background tag. This helps with negative class sampling during training to allow the model to learn from negative examples, which helps minimize false positives during inference. If assets with annotations are marked as Background, the annotations will be disregarded during training.