PCT Class Level
Click on Add Category, and select PCT to define the class.


Name
Export Name : Name of the class shown in the output data to be entered here.
Display Name : Class name displayed on the annotation tool to be entered here.
Abbreviated Name : Define the short form of the class name which gets displayed in data columns with space constraint.
Marking Tools
Once the PCT class has been added, you may pick what marking tools may be used with that class from the class settings as outlined above. Here are details regarding all of the marking tools available to use:
A 3D Cuboid to define the dimension and volume of an object. Suitable for, for example, a car.
This tool is affected by the Has Instances and by the Dynamic Dimensions properties in the PCT class settings.
A boundary with more than 3 sides around the points of an object in the point cloud. This can be suitable, for example, for buildings.

A line with a start and an end point with infinite joints/nodes. Suitable for, for example, lane markings.

A rectangle around an object in an image

A boundary with more that 3 sides around an object in an image.

A 2D line in an image.
Brush Sphere
Has Instance
Has Instance should be toggled ON when you need to track and differentiate between multiple objects of the same class within a frame. When enabled, each object instance is assigned a unique identifier for tracking and differentiation.
When to Enable
Enable Has Instance for classes where:
Multiple separate objects appear in the same frame
Individual objects need to be tracked across frames
Each object has distinct properties or behaviors
Example 'car_1', 'car_2', 'car_3'...
When to Disable
Disable Has Instance for classes where:
Only one annotation is needed per frame
The object represents a continuous region or background element
Individual tracking is not meaningful
Example 'Vegetation' or 'Sky'
Add Classification
Each class can have properties/attributes/classification that provide additional information about annotated objects. Classificiation help capture specific characteristics or states of the objects being labeled.
Boolean
A Boolean attribute has only two possible values: true or false. This is useful for capturing binary states or yes/no properties of an object.
Examples: Is the vehicle parked? Is the door open? Is the object occluded?
When you create a Boolean attribute, a Default Value toggle appears at the end of the form. This determines the initial state of the attribute when annotators create new annotations:
Toggle OFF: The attribute defaults to false in the annotation tool.
Toggle ON: The attribute defaults to true in the annotation tool.
Tip: Set the default to the most common state to speed up annotation. For example, if most vehicles are moving, set "Is Parked" to default false.
Single-Select Dropdown
A Single-Select attribute allows annotators to choose one option from a predefined list of values. This is ideal when an object can have only one state or category from multiple possibilities.
Examples: Vehicle type (sedan, SUV, truck), Traffic light color (red, yellow, green), Weather condition (sunny, cloudy, rainy)
Tip: Keep export consistent and lowercase for easier data processing. Use clear, descriptive display names that annotators will easily understand.
Multi-Select Dropdown
A Multi-Select attribute allows annotators to choose multiple options from a predefined list of values. This is ideal when an object can have several simultaneous states or characteristics.
Examples: Vehicle damage types (scratched, dented, broken glass), Visible features (headlights, license plate, mirrors), Weather conditions (foggy, rainy, windy)
Tip: Choose options that are mutually compatible. Unlike Single Select, annotators can select multiple options simultaneously for the same object.
Text
A Text attribute allows annotators to enter custom values in a text box rather than selecting from predefined options. This is useful when the attribute value is unique, variable, or cannot be predetermined.
Examples: License plate number, Vehicle identification number (VIN), Object ID, Custom notes
Settings
For all classification types (Boolean, Single Select, Multi-Select, Text), you must define the available options using three fields:
Export Name: The internal identifier stored in the output data (e.g.,
scratched,dented,broken_glass)Use lowercase, no spaces, and underscores for consistency
This is what appears in exported data and API responses
Display Name: The full display name shown to annotators in the annotation tool (e.g.,
Scratched,Dented,Broken Glass)Use clear, descriptive labels that are easy to understand
This is what annotators see and interact with
Abbreviated Name: A short abbreviation used when display space is limited (e.g.,
SCR,DNT,BRK)Keep it 2-4 characters when possible
Used in compact views and summarized displays
Associated Marking Tools
Select which annotation tools this attribute should be associated with. The attribute will only appear when annotators use the selected tool(s).
How It Works:
When you associate an attribute with specific marking tools, it will only be available for annotations created with those tools. This ensures annotators see only relevant attributes for each annotation type.
Attribute Mode
Specify how an attribute's value behaves over time and across sensors. This determines when annotators need to update attribute values.
Constant
The attribute value remains the same across all frames and sensors in the sequence.
Example: Vehicle color (a car's color doesn't change throughout the sequence)
Tip: Set once, applies everywhere
Time-Varying
The attribute value remains constant within each frame but can change across frames.
Example: Headlight status (lights can turn on/off between frames, but all sensors see the same state in each frame)
Tip: Same within frame, can change between frames
Sensor Specific
The attribute value can change both within a frame (different per sensor) and across frames.
Example: Occlusion level (an object may be partially hidden in one camera view but fully visible in another)
Tip: Independent value for each sensor in each frame
Last updated