Instructions
This tool leverages deep learning capabilities to perform instance segmentation on point clouds in power scenarios, based on a power scene point cloud classification model. After instance segmentation, individual instances such as power lines and poles in the power scene point cloud are obtained. Points belonging to the same instance share the same instance ID, which is recorded in the InstanceID attribute field of the result file.
Additionally, points of specified categories (e.g., power lines, poles) can be selected to participate in the instance segmentation calculation, thereby effectively improving segmentation efficiency and reducing the demand for computational resources.
Before using this tool, you need to obtain the resource extension package and configure the relevant environment. For detailed operations, please refer to the 3D AI Function Configuration Environment documentation.
Parameter Description
| Parameter Name | Description | Parameter Type |
|---|---|---|
| Point Cloud File Directory | Specifies the storage path of the source point cloud data. Currently, only *.las format is supported. | String |
| Retained Categories | Sets the category names and numbers to be retained. Only points of retained categories will participate in the instance segmentation calculation. Categories and numbers include: Power Line (number 14), Pole (number 15). | String |
| Storage Path | Sets the save path for the result point cloud data (*.las). | String |
Output Results
| Parameter Name | Description | Parameter Type |
|---|---|---|
| Result Path | Save path of the result point cloud data (*.las) | String |