class torch_points3d.core.data_transform.PlanarityFilter(thresh=0.3, is_leq=True)[source]

compute planarity and return false if the planarity of a pointcloud is above or below a threshold

  • thresh (float, optional) – threshold to filter low planar pointcloud

  • is_leq (bool, optional) – choose whether planarity should be lesser or equal than the threshold or greater than the threshold.

class torch_points3d.core.data_transform.RandomFilter(thresh=0.3)[source]

Randomly select an elem of the dataset (to have smaller dataset) with a bernouilli distribution of parameter thresh.


thresh (float, optional) – the parameter of the bernouilli function

class torch_points3d.core.data_transform.FCompose(list_filter, boolean_operation=numpy.logical_and)[source]

allow to compose different filters using the boolean operation

  • list_filter (list) – list of different filter functions we want to apply

  • boolean_operation (function, optional) – boolean function to compose the filter (take a pair and return a boolean)