Point Cloud Instance Segmentation

Feature Description

Supports classification and instance segmentation of point clouds in power scenarios. Based on deep learning capabilities, it can segment individual instances such as power lines and towers in power scenario point clouds. Points belonging to the same instance share identical instance IDs.

Additionally, supports excluding points of specified categories (e.g., ground points, vegetation points) from participating in instance segmentation calculations, thereby effectively improving segmentation efficiency and reducing computational resource requirements.

Before using this tool, users must obtain the extension resource package and configure the environment according to the included instructions. The extension package can be downloaded from: SuperMap Official Website->Product Download page->SuperMap iDesktopX specific version->Resource package containing Extension_AI 3D in its name.

This tool requires a Machine Learning Extension module license.

Parameter Description

Parameter Description Type
Point Cloud File Directory Source point cloud data storage path (currently only supports *.las format) String
Excluded Categories Specifies excluded category names and codes. E.g., Ground (ASPRS code: 2), Vegetation (ASPRS codes: 3, 4 or 5), Undefined (ASPRS codes: 0 or 1).
Points from excluded categories will not participate in instance segmentation.
String
Output Storage Directory Storage path for result point cloud data (*.las) String

Output

Parameter Description Type
Point Cloud Instance Segmentation Indicates whether instance segmentation succeeded Boolean