radiate module¶

class radiate.Sequence(sequence_path, config_file='config/config.yaml')¶

Bases: object

This class loads the sequence of RADIATE dataset

Example:
>>> import radiate
>>> root_path = ‘path/to/radiate/city_3_7/’
>>> seq = radiate.Sequence(root_path)
>>> output = seq.get_from_timestamp(seq.init_timestamp)
>>> seq.vis_all(output)

cfar(x, num_train, num_guard, rate_fa)¶

Detect peaks with CFAR algorithm.

Parameters

x (np.array) – input 1d array
num_train (int) – Number of training cells.
num_guard (int) – Number of guard cells.
rate_fa (float) – False alarm rate.

Returns

detected points

Return type

np.array

cfar2d(x, num_train, num_guard, rate_fa)¶

Detect peaks with 2D CFAR algorithm in each row.

Parameters

x (np.array) – input 2d array
num_train (int) – Number of training cells.
num_guard (int) – Number of guard cells.
rate_fa (float) – False alarm rate.

Returns

detected points

Return type

np.array

draw_boundingbox_rot(im, bbox, angle, color)¶

get_annotation_from_id(annotation_id)¶

get the annotation from an id

Parameters: annotation_id (int) – frame id
Returns: list of annotations for the id given as parameter
Return type: list

get_from_timestamp(t, get_sensors=True, get_annotations=True)¶

method to get sensor and annotation information from some timestamp

Parameters

t (float) – This is the timestamp which access the sensors/annotations
get_sensors (bool, optional) – whether to retrieve sensor information, defaults to True
get_annotations (bool, optional) – whether to retrieve annotation info, defaults to True

Returns

returns a single variable as a dictionary with ‘sensors’ and ‘annotations’ as key

Return type

dict

get_id(t, all_timestamps, time_offset=0.0)¶

get the closest id given the timestamp

Parameters

t (float) – timestamp in seconds
all_timestamps (np.array) – a list with all timestamps
time_offset (float, optional) – offset in case there is some unsynchronoised sensor, defaults to 0.0

Returns

the closest id

Return type

int

get_lidar_annotations(id_radar, interp=False, t_c=None, t_r1=None, t_r2=None)¶

get the annotations in lidar image coordinate frame

Parameters

id_radar (int) – the annotation radar id
interp (bool) – whether to use interpolation or not
t (float) – timestamp

Returns

the annotations in lidar image coordinate frame

Return type

dict

get_rectfied(left_im, right_im)¶

get the left and right image rectfied

Parameters

left_im (np.array) – raw left image
right_im (np.array) – raw right image

Returns

tuple (left_rect, right_rect, disp_to_depth) WHERE np.array left_rect is the rectfied left image np.array right_rect is the rectfied right image np.array disp_to_depth is a matrix that converts the disparity values to distance in meters

Return type

tuple

lidar_to_image(lidar)¶

Convert an lidar point cloud to an 2d bird’s eye view image

Parameters: lidar (np.array) – lidar point cloud Nx5 (x,y,z, intensity, ring)
Returns: 2d bird’s eye image with the lidar information
Return type: np.array

load_timestamp(timestamp_path)¶

load all timestamps from a sensor

Parameters: timestamp_path (string) – path to text file with all timestamps
Returns: list of all timestamps
Return type: dict

overlay_camera_lidar(camera, lidar)¶

Method that joins camera and projected lidar in one image for visualisation

Parameters

camera (np.array) – camera image
lidar (np.array) – lidar image with the same size as camera

Returns

overlayed image

Return type

np.array

project_bboxes_to_camera(annotations, intrinsict, extrinsic)¶

method to project the bounding boxes to the camera

Parameters

annotations (list) – the annotations for the current frame
intrinsict (np.array) – intrisic camera parameters
extrinsic (np.array) – extrinsic parameters

Returns

dictionary with the list of bbounding boxes with camera coordinate frames

Return type

dict

project_lidar(lidar, lidar_extrinsics, cam_intrinsic, color_mode='same')¶

Method to project the lidar into the camera

Parameters

lidar (np.array) – lidar point cloud with shape Nx5 (x,y,z,intensity,ring)
lidar_extrinsics (np.array) – 4x4 matrix with lidar extrinsic parameters (Rotation and translations)
cam_intrinsic (np.array) – 3x3 matrix with camera intrinsic parameters in the form [[fx 0 cx], [0 fx cy], [0 0 1]]
color_mode (string) – what type of information is going to be representend in the lidar image

options: ‘same’ always constant color. ‘pseudo_distance’: uses a color map to create a psedo color which refers to the distance. ‘distance’ creates an image with the actual distance as float

Return type: np.array
Returns: returns the projected lidar into the respective camera with the same size as the camera

read_lidar(lidar_path)¶

given a lidar raw path returns it lidar point cloud

Parameters: lidar_path (string) – path to lidar raw point
Returns: lidar point cloud Nx5 (x,y,z,intensity,ring)
Return type: np.array

transform_annotations(annotations, M)¶

method to transform the annotations to annother coordinate

Parameters

annotations (list) – the list of annotations
M (np.array) – transformation matrix

Returns

the list of annotations in another coodinate frame

Return type

list

transform_point_cloud(pc, M)¶

transform a 3d point cloud to another coordinate frame

Parameters

pc (np.array) – point cloud in the form Nx% (x,y,z,intensity, ring)
M (np.array) – transformation matrix

Returns

transformed point cloud

Return type

np.array

vis(sensor, objects, color=None, mode='rot')¶

visualise the sensor and its annotation

Parameters

sensor (the given sensor) –
objects (list of objects) – np.array

Returns

image with the objects overlayed

Return type

np.array

vis_3d_bbox_cam(image, bboxes_3d, pc_size=0.7)¶

diplay pseudo 3d bounding box from camera

Parameters

image (np.array) – camera which the bounding box is going to be projected
bboxes_3d (dict) – list of bounding box information with pseudo-3d image coordinate frame
pc_size (float) – percentage of the size of the bounding box [0.0 1.0]

Returns

camera image with the correspondent bounding boxes

Return type

np.array

vis_all(output, wait_time=1)¶

method to diplay all the sensors/annotations

Parameters

output (dict) – gets the output from self.get_from_timestamp(t)
wait_time (int, optional) – how to long to wait until display next frame. 0 means it will wait for any key, defaults to 1

vis_bbox_cam(image, bboxes_3d, pc_size=0.7)¶

diplay pseudo 2d bounding box from camera

Parameters

image (np.array) – camera which the bounding box is going to be projected
bboxes_3d (dict) – list of bounding box information with pseudo-3d image coordinate frame
pc_size (float) – percentage of the size of the bounding box [0.0 1.0]

Returns

camera image with the correspondent bounding boxes

Return type

np.array