1. Basic Details

Fill in the basic details about the labeling project:

1. Annotation Tool

2. Name and Description

3. Sequential Task

4. Reference Images

5. Overlapping of Annotations

6. Tracking in Multi-Sensor

Annotation Tool

Select the Annotation Tool based on the type of labeling requirement. The default annotation tool selected is ‘iMerit LiDAR (Multisensor)’

Name and Description

Give the recipe a name and description for contributors to understand.

Sequential Task

Toggle this option ON if the uploaded frames belong to a single batch. Enabling this will activate the keyframe interpolation feature in the tool, allowing seamless interpolation across frames. Leaving this toggle button ON for single-frame batches has no adverse effect on the annotator's experience.

Assign all frames of a sequence to a single annotator. This ensures they have object context from adjacent frames, aiding in identifying objects that may be unclear in the current frame.

Reference Images

Toggle this option ON if the task requires referencing camera images for object identification or labeling. Enabling this will display an image panel within the annotation tool for easy reference.

Calibration files of images will be needed at the time of data import. Learn how to prepare the calibration file.

Overlapping of Annotations

Toggle this option ON to allow two annotations to overlap without conflict detection. Turning it OFF will enable the tool to identify overlapping annotations and display a conflict alert. When a conflict is detected, the user will have the option to override the first classification at the overlapping instance.

Tracking in Multi-Sensor

Enable this option to track an object across multiple sensors within a frame or across frames in a batch. When toggled ON, the tool will assign a unique ID to the object, allowing the identifier to be shared across visible sensors seamlessly.

Segmentation Mode

Toggle this option ON to enable the segmentation tool. This will disable the geometry tool and allow annotators to segment points within the point cloud and pixels in the camera sensor images.

Relationship Sensor Dependent

• Toggle this option ON to allow relationships to be defined independently within each sensor view. This means the same objects can have different relationships (or none at all) depending on the sensor perspective, giving you more granular control in multi-sensor scenarios. Toggle it OFF to create frame-level relationships that automatically apply across all sensor views.

• By default, the tool supports the toggle ON option.

Learn how to use