vis4d.data.transforms.points

Pointwise transformations.

Classes

AddGaussianNoise(*[, in_keys, out_keys, ...])	Adds random normal distributed noise with given std to the data.
AddUniformNoise(*[, in_keys, out_keys, ...])	Adds random normal distributed noise with given std to the data.
ApplySE3Transform(*[, in_keys, out_keys, ...])	Applies a given SE3 Transform to the data.
ApplySO3Transform(*[, in_keys, out_keys, ...])	Applies a given SO3 Transform to the data.
CenterAndNormalize(*[, in_keys, out_keys, ...])	Centers and normalizes the pointcloud.
GenPcBounds(*[, in_keys, out_keys, sensors, ...])	Extracts the max and min values of the loaded points.
NormalizeByMaxBounds(*[, in_keys, out_keys, ...])	Normalizes the pointcloud by the max bounds.
SE3Transform	Parameters for Resize.
TransposeChannels(*[, in_keys, out_keys, ...])	Transposes some predifined channels.

class AddGaussianNoise(*, in_keys=['points3d'], out_keys=['points3d'], sensors=None, same_on_batch=True, **kwargs)

Adds random normal distributed noise with given std to the data.

Parameters:

std (float) – Standard Deviation of the noise

__call__(coordinates_list)

Adds gaussian noise to the coordiantes.

Return type:

list[Union[ndarray[Any, dtype[float32]], ndarray[Any, dtype[float64]]]]

class AddUniformNoise(*, in_keys=['points3d'], out_keys=['points3d'], sensors=None, same_on_batch=True, **kwargs)

Adds random normal distributed noise with given std to the data.

Parameters:

std (float) – Standard Deviation of the noise

__call__(coordinates_list)

Adds uniform noise to the coordinates.

Return type:

list[Union[ndarray[Any, dtype[float32]], ndarray[Any, dtype[float64]]]]

class ApplySE3Transform(*, in_keys=['points3d'], out_keys=['points3d'], sensors=None, same_on_batch=True, **kwargs)

Applies a given SE3 Transform to the data.

__call__(coordinates_list)

Applies a SE3 Transform.

Return type:

list[Union[ndarray[Any, dtype[float32]], ndarray[Any, dtype[float64]]]]

class ApplySO3Transform(*, in_keys=['points3d'], out_keys=['points3d'], sensors=None, same_on_batch=True, **kwargs)

Applies a given SO3 Transform to the data.

__call__(coordinates_list)

Applies a given SO3 Transform to the data.

Return type:

list[Union[ndarray[Any, dtype[float32]], ndarray[Any, dtype[float64]]]]

class CenterAndNormalize(*, in_keys=['points3d'], out_keys=['points3d'], sensors=None, same_on_batch=True, **kwargs)

Centers and normalizes the pointcloud.

__call__(coords_list)

Applies the Center and Normalization operations.

Return type:

list[Union[ndarray[Any, dtype[float32]], ndarray[Any, dtype[float64]]]]

class GenPcBounds(*, in_keys=['points3d'], out_keys=['transforms.pc_bounds'], sensors=None, same_on_batch=True, **kwargs)

Extracts the max and min values of the loaded points.

__call__(coordinates_list)

Extracts the max and min values of the pointcloud.

Return type:

list[Union[ndarray[Any, dtype[float32]], ndarray[Any, dtype[float64]]]]

class NormalizeByMaxBounds(*, in_keys=('points3d', 'trasforms.pc_bounds'), out_keys=['points3d'], sensors=None, same_on_batch=True, **kwargs)

Normalizes the pointcloud by the max bounds.

__call__(coords_list, pc_bounds_list)

Applies the normalization.

Return type:

list[Union[ndarray[Any, dtype[float32]], ndarray[Any, dtype[float64]]]]

class SE3Transform

Parameters for Resize.

class TransposeChannels(*, in_keys=['points3d'], out_keys=['points3d'], sensors=None, same_on_batch=True, **kwargs)

Transposes some predifined channels.

__call__(coordinates_list)

Transposes some predifined channels.

Return type:

list[Union[ndarray[Any, dtype[float32]], ndarray[Any, dtype[float64]]]]