Release Notes - Multi Point Cloud
Updated at June 7th, 2023
👍 More Points, more precision
The Sama workspace now supports the ability to annotate point cloud data coming from multiple detection sensors.
To address increasingly complex 3D Point Cloud and Sensor Fusion workloads, annotation projects can now be configured with multiple points clouds in a single 3D output.
How it works
With Sama's previous Point Cloud annotation, all points in the cloud would come from a single defined sensor. As a result, there were many cases where details could be obscured or objects could be difficult to see. Now, you're able to provide Point Cloud Data (PCD) for multiple sensors, resulting in increased data accuracy in the workspace.
Each pcd file will contain details about the sensor itself, and cuboids added to the workspace will be assigned a lead sensor based on the points used in its creation. This lead sensor is treated as a dynamic nested output, and over the object's life cycle can change from one lead sensor to another. Points from each sensor will be colored differently to facilitate identification.
👍Tips
Think of the lead sensor as the strongest sensor at the time - whichever points are providing the clearest image could be assigned as the lead sensor.
In addition to the lead sensor addition, a new window has been added to the workspace. This window allows annotators to identify all sensors in the point cloud. This window can also be used to toggle the visibility of certain sensors, enabling only specific points to show in the workspace.
What you need
To deploy a Multi Point Cloud project, you need to organize your data folders in a new way. Multi Point Cloud folders must contain not only the initial point cloud folder, but a subfolder for each LIDAR sensor.
Multi Point Cloud projects also require additional project configuration. For more information, reach out to your Sama Solutions Engineer or Sama Delivery Manager.