Pular para o conteúdo principal

Contextualize point cloud

You can contextualize point clouds in two ways: by manually creating each object you want to contextualize or by leveraging the point cloud's segmentation. You must use both services to achieve a fully contextualized point cloud.

Before you start

Make sure you have already uploaded the point cloud you want to contextualize. For more information on uploading point clouds, see Upload point cloud.

The following section will explain how to contextualize point clouds.

Get started

Before contextualizing, you should filter and index the point cloud to generate tiles that you can use for object detection and accelerate the contextualization process.

It is an automated process triggered by selecting the filter icon in your model's action column.

  1. Under Actions, select ... and Filtering and indexing icon Filtering and indexing.

  2. Select the check box to generate tiles.

  3. Select OK and the process starts.

Once the filtering and indexing are processed, trigger the object detection. The Run classification Run classification icon icon will now be enabled.

  1. Select Run classification to start the point cloud object detection.

  2. Enable both Classification and Pipe detection.

  3. Select Run.

Classification will perform semantic segmentation of the point cloud, while pipe detection will detect segments in the point clouds that resemble pipes.

Once an object is detected, you will see an orange tag on the right, suggested annotations indicating how many objects were detected.


Once the filtering and detecting objects are ready, select Contextualize to start the contextualization process. This opens the point cloud, and you can contextualize within the 3D environment window.

Tools in contextualization

In the top navigation bar, there is a drop-down menu for all the suggestions. These are suggestions for classifications and pipes. After contextualizing, you will see a drop-down for all the contextualized objects.

Use slicing Slice icon to get a better overview of the model you are contextualizing. For example, you can slice your viewpoint below ceiling height.

  • You can turn off and on the highlights of the different suggestions of object types from the drop-down. To focus on contextualizing one object type at a time, select Toggle annotation visibility Toggle annotation visibility icon in the toolbar on the left to turn off the highlight for all other suggestions.

  • You can also color the model by classification. Use the filter Filter icon icon from the left toolbar. Enable View classification to view all or a selection of the classifications.

  • Select Fit to view Fit to view icon to reorient yourself and reposition the camera view outside the model.

  • When you start the contextualization, you will see that the cursor Cursor icon icon is activated and blue. Use Select Annotation to select any detected object and annotations.

  • Select Add annotation icon (Add annotation) to add an annotation in the point cloud.

  • After adding an annotation, in the resource selector, use the Filter icon Filters to refine the results. Use  to sort and select columns. You can sort by name, object type, location and more. See Search for more information.

  • To zoom an annotation, select Zoom to annotation.

  • Use Settings Settings icon icon to adjust the coloring of the points in the model.

Contextualize automatically detected objects

To link an object to another asset:

  1. To find the object you want to contextualize, navigate to Suggestions in the top bar and select an object.

  2. In the resource selector on the top right, select Relink.

  3. Use the search bar in the resource selector to find the asset you want to map to the selected object.

  4. Select the asset, verify it is correct, and select + Add.

You have contextualized the object. The annotation count and contextualization count show how many objects you annotated.

Contextualize undetected objects

Point clouds are sets of data points in a three-dimensional coordinate system. Each point in the cloud represents a specific position in space and can contain additional information like color or intensity. This means a point is not related to nearby points, and some objects might not be recognized. To contextualize a group of points as an object, you draw a shape around the points and contextualize that shape.

To contextualize an undetected object:

Use the Add annotation icon icon to add an annotation in the point cloud.

  1. Select Add annotation icon (Add annotation) from the toolbar on the left to add an annotation.

  2. In the model, select where you want to contextualize the object.

  3. A blue cube will appear in the 3D view. Adjust the cube to fit the object. Use either:

    1. Translate (T) will move the cube in the x, y, and z directions. (or)

    2. Rotate (R) will rotate the cube.

    3. Scale (G) will scale the size of the box.

  4. After creating the shape, you can use the resource selector on the right to map an asset to the selected object. Use the search bar to find an existing asset.

  5. Select the asset, verify it is correct, and select + Add.

  6. Select Create annotation in the toolbar at the bottom of the screen.

Update or delete annotation

In point clouds, you have a 1:1 relationship between assets and objects. This means you can only link one asset to a detected object. You can relink, update, or delete an annotation.

To relink the object to a different asset or to update the annotation:

  1. Use the cursor tool and select the annotated object.

  2. Select Relink in the top right of the resource selector.

  3. Search for the asset you want to map to the selected object.

  4. Select + Add.

You have updated the annotation.

To delete the annotation using Annotations:

  1. Open the Annotations dropdown.

  2. Search for the asset you want to delete. Hover over each annotation to see the highlighted objects.

  3. Select Delete icon to delete the annotation.

You have deleted the annotation.