Before importing/making annotations we first need to create the needed annotation classes, such as a polygon cell
or a bounding box car
.
Every annotation class has exactly one main type (polygon, bounding box etc) and zero or more sub types (attributes, text etc).
Classes are share across the whole team, this requires the name to be unique.
Each class needs to be associated with every dataset that wants to use them. This is done by adding the dataset id/slug to the datasets
field in Update Annotation Class.
In the table below are each annotation type listed, note that the api uses the slug instead of the name.
Name | Slug | Type |
---|---|---|
Polygon | polygon | Main |
Bounding Box | bounding_box | Main |
Tag | tag | Main |
Skeleton | skeleton | Main |
Keypoint | keypoint | Main |
Line | line | Main |
Ellipse | ellipse | Main |
Cuboid | cuboid | Main |
Attributes | attributes | Sub |
Text | text | Sub |
Directional Vector | directional_vector | Sub |
Instance ID | instance_id | Sub |