lidar_aug.python package¶

This package contains utility code, Python wrappers for point cloud transformations as well as a class to visualize point clouds.

Submodules¶

lidar_aug.python.augmentations module¶

This submodule contains Python wrappers for many transformations. These wrappers do a shape check and, in some instances, do some pre- / post-processing of inputs/outputs.

lidar_aug.python.augmentations.apply_transformation(points: Tensor, transformation_matrix: Tensor) → None¶

Applies a transformation matrix to an entire point cloud with the shape (B, N, F), where B is the number of batches and N is the number of points.

Parameters:

points – is the point cloud that the transformation matrix is applied to. Expected shape is (B, N, 4).
transformation_matrix – is the transformation matrix.

lidar_aug.python.augmentations.delete_labels_by_min_points(points: Tensor, labels: Tensor, names: Tensor, min_points: int) → None¶

Checks the amount of points for each bounding box. If the number of points is smaller than a given threshold, the box is removed along with its label.

Parameters:

points – is the point_cloud. Expected shape is (B, N, 4).
labels – are the bounding boxes of objects. Expected shape is (B, N, 7).
names – are the names/labels of these boxes.
min_points – is the point threshold.

lidar_aug.python.augmentations.flip_random(points: Tensor, labels: Tensor, prob: int) → None¶

Flips all the points in the point cloud with a probability of prob % in the direction of the y-axis.

Parameters:

points – is the point cloud containing the points that will be flipped. Expected shape is (B, N, 4).
labels – are the corresponding labels. Expected shape is (B, N, 7).
prob – is the probability with which the points should be flipped.

lidar_aug.python.augmentations.intensity_noise(points: Tensor, sigma: float, max_intensity: IntensityRange) → None¶

Shifts the intensity value of every point in the point cloud by a random amount drawn from a normal distribution.

Parameters:

points – is the point cloud with all the points. Expected shape is (B, N, 4).
sigma – is the standard deviation of the normal distribution.
max_intensity – is the maximum intensity value (either 1 or 255, depending on the dataset).

lidar_aug.python.augmentations.intensity_shift(points: Tensor, sigma: float, max_intensity: IntensityRange) → None¶

Shifts the intensity value of every point in the point cloud by a single value drawn from a normal distribution.

Parameters:

points – is the point cloud with all the points. Expected shape is (B, N, 4).
sigma – is the standard deviation of the normal distribution.
max_intensity – is the maximum intensity value (either 1 or 255, depending on the dataset).

lidar_aug.python.augmentations.local_to_local_transform(from_pose: Tensor, to_pose: Tensor) → Tensor¶

Creates a transformation matrix from the local system into a ‘target’ coordinate frame.

Parameters:

from_pose – is the local coordinate frame (x, y, z, roll, yaw, pitch).
to_pose – is the target coordinate frame (x, y, z, roll, yaw, pitch).

Returns:

the homogeneous transformation matrix into the target coordinate frame.

lidar_aug.python.augmentations.local_to_world_transform(lidar_pose: Tensor) → Tensor¶

Creates a transformation matrix from the local system into the global coordinate frame.

Parameters:: lidar_pose – is the local coordinate frame (x, y, z, roll, yaw, pitch).
Returns:: the homogeneous transformation matrix into the global coordinate frame.

lidar_aug.python.augmentations.random_noise(points: Tensor, sigma: float, ranges: list[float] | DistributionRanges, noise_type: NoiseType, max_intensity: IntensityRange) → None¶

Adds random amount of points (drawn using a normal distribution) at random coordinates (within predetermined ranges) with a random intensity according to specific noise type.

Parameters:

points – is the point cloud that the points will be added to. Expected shape is (B, N, 4).
sigma – is the standard deviation of the normal distribution that is used to draw the number of points to be added.
ranges – are the boundaries in (min and max (x, y, z) values) in which the new points can be created.
noise_type – is one of a number of ‘patterns’ that can be used to generate the points.
max_intensity – is the maximum intensity value in the dataset

lidar_aug.python.augmentations.random_point_noise(points: Tensor, sigma: float) → None¶

Moves each point in the point cloud randomly. How much each coordinate is changed is decided by values drawn from a normal distribution.

Parameters:

points – is the point cloud from which each point is moved. Expected shape is (B, N, 4).
sigma – is the standard deviation of the normal distribution.

lidar_aug.python.augmentations.rotate_deg(points: Tensor, angle: float) → None¶

Rotates a batch of points along the ‘z’ axis (yaw).

Parameters:

points – is the point cloud that the rotation is applied to. Expected shape is (B, N, 4).
angle – is the angle (in degrees) by which the points are to be rotated.

lidar_aug.python.augmentations.rotate_rad(points: Tensor, angle: float) → None¶

Rotates a batch of points along the ‘z’ axis (yaw).

Parameters:

points – is the point cloud that the rotation is applied to. Expected shape is (B, N, 4).
angle – is the angle (in radians) by which the points are to be rotated.

lidar_aug.python.augmentations.rotate_random(points: Tensor, labels: Tensor, sigma: float) → None¶

Rotates points and labels. The number of degrees that they are rotated by is determined by a randomly generated value from a normal distribution.

Parameters:

points – is the point cloud that the rotation is applied to. Expected shape is (B, N, 4).
labels – are the labels belonging to the point cloud that the rotation is applied to. Expected shape is (B, N, 7).
sigma – is the standard deviation of the normal distribution.

lidar_aug.python.augmentations.scale(points: Tensor, scaling_factor: float) → None¶

Scales points by a constant factor.

Parameters:

points – is the point cloud whose points are scaled. Expected shape is (B, N, 4).
scaling_factor – is the factor that the (x, y, z) coordinates are multiplied by.

lidar_aug.python.augmentations.scale_local(points: Tensor, labels: Tensor, sigma: float, max_scale: float) → None¶

Scales the points that are part of a box and the corresponding labels by a random factor.

This factor is drawn from a truncated normal distribution. The truncated normal distribution has a mean of 1. The standard deviation, as well as upper and lower limits are determined by the function parameters.

Parameters:

points – is the point cloud that contains the points that will be scaled. Expected shape is (B, N, 4).
labels – are the labels belonging to the aforementioned point cloud. Expected shape is (B, N, 7).
sigma – is the standard deviation of the truncated normal distribution.
max_scale – is the upper limit of the truncated normal distribution. The lower limit is the inverse.

lidar_aug.python.augmentations.scale_random(points: Tensor, labels: Tensor, sigma: float, max_scale: float) → None¶

Scales the points and labels by a random factor. This factor is drawn from a truncated normal distribution. The truncated normal distribution has a mean of 1. The standard deviation, as well as upper and lower limits are determined by the function parameters.

Parameters:

points – is the point cloud that contains the points that will be scaled. Expected shape is (B, N, 4).
labels – are the labels belonging to the aforementioned point cloud. Expected shape is (B, N, 7).
sigma – is the standard deviation of the truncated normal distribution.
max_scale – is the upper limit of the truncated normal distribution. The lower limit is the inverse.

lidar_aug.python.augmentations.thin_out(points: Tensor, sigma: float) → None¶

Randomly generates a percentage from a normal distribution, which determines how many items should be ‘thinned out’. From that percentage random indices are uniformly drawn (in a random order, where each index is unique).

Finally, a new tensor is created containing the items present at those indices.

Parameters:

points – is the point cloud. Expected shape is (B, N, 4).
sigma – is the standard deviation of the distribution that generates the percentage.

lidar_aug.python.augmentations.transform_along_ray(points: Tensor, sigma: float) → None¶

Moves each point in the point cloud randomly along a ray. How much it is moved is decided by a value drawn from a normal distribution.

Parameters:

points – is the point cloud from which each point is moved. Expected shape is (B, N, 4).
sigma – is the standard deviation of the normal distribution.

lidar_aug.python.augmentations.translate(points: Tensor, translation: Tensor) → None¶

Moves points by a specific amount.

Parameters:

points – is the point cloud with the points are to be moved. Expected shape is (B, N, 4).
translation – is the translation vector that specifies by how much they points are moved.

lidar_aug.python.augmentations.translate_random(points: Tensor, labels: Tensor, sigma: float) → None¶

Generates a random (3D) translation vector using a normal distribution and applies it to all the points and labels.

Parameters:

points – is the point cloud with the points that are translated. Expected shape is (B, N, 4).
labels – are the labels belonging to the aforementioned point cloud. Expected shape is (B, N, 7).
sigma – is the standard deviation of the normal distribution.

lidar_aug.python.utils module¶

This submodule contains various utility functions.

class lidar_aug.python.utils.Calibration(calib_file)¶

Bases: object

cart_to_hom(pts)¶

Parameters:: pts – (N, 3 or 2)
Return pts_hom:: (N, 4 or 3)

corners3d_to_img_boxes(corners3d)¶

Parameters:: corners3d – (N, 8, 3) corners in rect coordinate
Returns:: boxes: (None, 4) [x1, y1, x2, y2] in rgb coordinate
Returns:: boxes_corner: (None, 8) [xi, yi] in rgb coordinate

img_to_rect(u, v, depth_rect)¶

Parameters:

u –
v –
depth_rect –

Returns:

lidar_to_img(pts_lidar)¶

Parameters:: pts_lidar – (N, 3)
Return pts_img:: (N, 2)

lidar_to_rect(pts_lidar)¶

Parameters:: pts_lidar – (N, 3)
Return pts_rect:: (N, 3)

rect_to_img(pts_rect)¶

Parameters:: pts_rect – (N, 3)
Return pts_img:: (N, 2)

rect_to_lidar(pts_rect)¶

Parameters:: pts_lidar – (N, 3)
Return pts_rect:: (N, 3)

class lidar_aug.python.utils.Object3d(line)¶

Bases: object

cam_to_velo(calib)¶

generate_corners3d()¶

get_lidar_bbox(calib)¶

lidar_aug.python.utils.boxes_to_corners_3d(boxes3d)¶

Creates a box representation using its corners like this:

  7 -------- 4
 /|         /|
6 -------- 5 .
| |        | |
. 3 -------- 0
|/         |/
2 -------- 1

Parameters:: boxes3d – (N, 7) [x, y, z, dx, dy, dz, heading], (x, y, z) is the box center
Returns:: a new tensor representing the box by its eight corners

lidar_aug.python.utils.check_numpy_to_torch(x)¶

lidar_aug.python.utils.create_distribution_ranges(input: list) → DistributionRanges¶

lidar_aug.python.utils.get_calib_from_file(calib_file)¶

lidar_aug.python.utils.get_objects_from_label(label_file)¶

lidar_aug.python.utils.gt_boxes_from_label(label_file, calib)¶

lidar_aug.python.utils.in_hull(p, hull)¶

Parameters:

p – (N, K) test points
hull – (M, K) M corners of a box

Returns:

bool

lidar_aug.python.utils.load_point_cloud_from_file(pc_path, point_features=4, normalize_intensities=False)¶

lidar_aug.python.utils.points_in_boxes_cpu(points, boxes)¶

Parameters:

points – (num_points, 3)
boxes – [x, y, z, dx, dy, dz, heading], (x, y, z) is the box center, each box DO NOT overlaps

Returns:

(N, num_points)

Return type:

point_indices

lidar_aug.python.visualization module¶

This submodule contains functions to visualize point clouds.

class lidar_aug.python.visualization.PCViewer(point_cloud=None, gt_boxes=None, pred_boxes=None, point_features=4, normalize_intensities=False, calib=None)¶

Bases: object

draw(normalize_intensities=False)¶

set_gt_boxes(gt_boxes)¶

set_gt_boxes_from_label(label_file, calib_file)¶

set_point_cloud(point_cloud)¶

set_pred_boxes(pred_boxes, pred_scores=None, min_score=0.5)¶

lidar_aug.python package¶

Submodules¶

lidar_aug.python.augmentations module¶

lidar_aug.python.utils module¶

lidar_aug.python.visualization module¶

Module contents¶

LidarAug

Navigation

Related Topics