Skip to content

Feature/label cloud from image segmentation

Lea Vauchier requested to merge feature/labelCloudFromImageSegmentation into master

Side effects:

  • Refactor BboxesFromImagesDetections

Add new StandAlone tools to:

  • label a point cloud given a panoptic segmentation on corresponding images (LabelCloudFromImagesDetections)
  • refine this segmentation using clues from the 3D cloud (PostProcessLabelCloud)

Create a SegmentedCloudTransformations module to work on segmented clouds

Expected Side effects:

  • BboxFromImagesDetections refactored to reuse common code
  • New vtk filter vtkSeparateCloudKnn
  • vtkBoundingBoxReader can now parse confidence and time from yaml file in vtkBoundingBoxReader
  • Cylindrical bounding volumes are used for "person" 3D bounding box visualisation
Edited by Lea Vauchier

Merge request reports