kitcar_ml.utils.data_generation package

Subpackages

• kitcar_ml.utils.data_generation.test package
  • Submodules
  • kitcar_ml.utils.data_generation.test.test_sample_gen module
    • test_position_gen()
    • test_tf_to_img()
    • main()
  • Module contents

Submodules

kitcar_ml.utils.data_generation.augmentation_utils module

Functions:

gaussian_noise(img, sigma, mean)	Add gaussian noise to image.
add_random_intensity_and_contrast(img, ...)	Add intensity/contrast to image.
adapt_contrast_and_mean(img, dev_bckg_img)	Adapt contrast/mean of img to background.
supersample(img, s)	Supersample (resize) given image.
revert_supersampling(img, s)	Revert previous supersampling (resize) of given image.
add_object_to_background(background_img, ...)	Blend artificial image into background image.

gaussian_noise(img: ndarray, sigma: float, mean: float) → ndarray

Add gaussian noise to image.

add_random_intensity_and_contrast(img: ndarray, sigma_all: float, mean_all: float, gauss_blur_all: float)

Add intensity/contrast to image.

adapt_contrast_and_mean(img: ndarray, dev_bckg_img: float)

Adapt contrast/mean of img to background.

supersample(img: ndarray, s: float) → ndarray

Supersample (resize) given image.

revert_supersampling(img: ndarray, s: float) → ndarray

Revert previous supersampling (resize) of given image.

add_object_to_background(background_img: ndarray, aug_img: ndarray, img_coords: ndarray, noise_sigma: float, noise_mean: float, gauss_blur_sign: int, bckg_img_stddev: float, hor_motion_blur_size: int) → ndarray

Blend artificial image into background image.

Parameters

• background_img – Image without the current generation.

• aug_img – Pattern that should be inserted into background image.

• img_coords – Coordinates where to insert the aug_img.

• noise_mean – Mean of the noise.

• noise_sigma – Stddev of the noise.

• gauss_blur_sign – Gaussian blur kernel size applied to aug_img.

• bckg_img_stddev – Stddev of noise applied to the background img.

• hor_motion_blur_size – Kernel size of motion blur applied in hor direction.

Returns

Background image with aug_img added to it.

kitcar_ml.utils.data_generation.data_generation_tool module

Functions:

run_tasks(func, input_images, num_workers)
main(image_dir[, config_path, ...])

run_tasks(func: Callable[[str], None], input_images: List[str], num_workers: int)

main(image_dir: str, config_path: Optional[str] = None, path_dataset_generated: str = '.', augmentation_files: str = './images_to_be_inserted', seed: int = 1, ts_sizes_path: Optional[str] = None, camera_specs_path: Optional[str] = None, oversampling_rate: int = 1, num_workers: int = 4)

kitcar_ml.utils.data_generation.generation_config module

Classes:

GenerationConfiguration(supersampling, ...)

class GenerationConfiguration(supersampling: int, max_number_of_objects: int, min_distance: float, max_distance: float, angle_x_max: float, angle_z_max: float, noise_sigma: float, noise_mean: float, gauss_blur_all: float, gauss_blur_sign: float, overlay_treshold: float, sigma_all: float, mean_all: float, crop_box: List[int], bbox_min_visibility: float, bbox_min_area: int, hor_motion_blur_size: float)

Bases: InitOptions

Attributes:

supersampling
max_number_of_objects
min_distance
max_distance
angle_x_max
angle_z_max
noise_sigma
noise_mean
gauss_blur_all
gauss_blur_sign
overlay_treshold
sigma_all
mean_all
crop_box
bbox_min_visibility
bbox_min_area
hor_motion_blur_size

supersampling: int

max_number_of_objects: int

min_distance: float

max_distance: float

angle_x_max: float

angle_z_max: float

noise_sigma: float

noise_mean: float

gauss_blur_all: float

gauss_blur_sign: float

overlay_treshold: float

sigma_all: float

mean_all: float

crop_box: List[int]

bbox_min_visibility: float

bbox_min_area: int

hor_motion_blur_size: float

kitcar_ml.utils.data_generation.object_generation_utils module

Functions:

calculate_overlay(bb_old, bb_new)
clean_bbox(x_coords, y_coords, crop_box, ...)	Sort/Clip coordinates and filter invalid/overlayed bboxes.
sample_objects(number_of_objects, name_list, ...)	Try to create randomized objects that have valid bboxes.
add_objects_to_image(max_number_of_objects, ...)	Add multiple objects to image.
create_artificial_objects_in_image(image, ...)	Add artificial objects to the given image.
load_and_create_artificial_images(img_names, ...)	Load provided images and generate synthetic images out of them.

Classes:

GeneratedObject(img_coords, aug_idx, label)

calculate_overlay(bb_old, bb_new)

clean_bbox(x_coords: Sequence[int], y_coords: Sequence[int], crop_box: Tuple[int, int, int, int], min_visibility, min_area, max_overlay, other_bboxes) → Optional[Tuple[int, int, int, int]]

Sort/Clip coordinates and filter invalid/overlayed bboxes.

class GeneratedObject(img_coords: Sequence[int], aug_idx: int, label: str)

Bases: object

Attributes:

img_coords
aug_idx
label

img_coords: Sequence[int]

aug_idx: int

label: str

sample_objects(number_of_objects: int, name_list: List[str], sample_generator: SampleGenerator, angle_x_max: float, angle_z_max: float, overlay_trshld: float, crop_box: float, bbox_min_visibility: float, bbox_min_area: int) → List[GeneratedObject]

Try to create randomized objects that have valid bboxes.

add_objects_to_image(max_number_of_objects: int, name_list: List[str], aug_images: List[ndarray], input_image: ndarray, sample_generator: SampleGenerator, supersampling: float, angle_x_max: float, angle_z_max: float, noise_sigma: float, noise_mean: float, gauss_blur_sign: float, overlay_trshld: float, crop_box: float, bbox_min_visibility: float, bbox_min_area: int, hor_motion_blur_size: int) → Tuple[ndarray, List[str]]

Add multiple objects to image.

Parameters

• max_number_of_objects – # of inserted objects.

• name_list – Names of available objects.

• aug_images – Available generation images.

• input_image – Image that should be altered.

• sample_generator – Generator of sample points for generations.

• supersampling – Supersampling factor.

• angle_x_max – Maximum rotation of generation around x axis.

• angle_z_max – Maximum rotation of generation in z axis (after x rotation).

• noise_sigma – Stddev of noise applied to generation image.

• noise_mean – Mean of noise applied to generation image.

• gauss_blur_sign – Filter kernel size of gaussian blur.

• overlay_trshld – Threshold of max. overlay between two generations.

• crop_box – ROI.

• min_pxl_size – Minimal size of generation in pixels.

• bbox_min_visibility – Minimal portion of generated bbox that is visible.

• bbox_min_area – Minimal area of generated bbox.

• hor_motion_blur_size – Kernel size of motion blur applied in hor direction.

Returns

Augmented image and list of labels.

create_artificial_objects_in_image(image: List[ndarray], config: GenerationConfiguration, name_list, aug_images, sample_generator) → Tuple[ndarray, List[str]]

Add artificial objects to the given image.

The objects are sampled from aug_images and added as labels to the dataset.

Parameters

• image – Image.

• config_data – Configurations; e.g. max number of objects in one image.

• name_list – Class names of objects.

• aug_images – Images of the objects (e.g. image of a stop sign)

• sample_generator – Generator of positions in vehicle coordinates.

• dataset – Dataset

• naming_convention – Callable that returns a unique name for each index.

Returns

Altered image and labels.

load_and_create_artificial_images(img_names: List[str], dataset, config, name_list, aug_images, sample_generator)

Load provided images and generate synthetic images out of them.

Parameters

• img_names – Names(paths) to all images that should be used.

• dataset – The dataset.

• config – Configuration of the image generation.

• name_list – Names of available generations.

• aug_images – Available generation images.

• sample_generator – Generator of sample points for generations.

kitcar_ml.utils.data_generation.sample_generator module

Classes:

SampleGenerator(camera_specs_path, ...)

class SampleGenerator(camera_specs_path, ts_sizes_path, min_distance, max_distance)

Bases: object

Attributes:

camera_specs	Camera specification including calibration.
ts_sizes	Sizes of all traffic signs.
min_distance	Minimal distance to generated sign.
max_distance	Maximal distance to generated sign.

Methods:

sample_world_vec()	Sample vector to point in the fov.
_H(height)	Calculate transform from vehicle coords to img at a given height.
vehicle_point_to_img(vec)	Transform vector in vehicle coords to pixel.
sign_coords(sign_name, angle_x, angle_z)	Calculate rotated sign coordinates.

camera_specs: CameraSpecs

Camera specification including calibration.

ts_sizes: Dict[str, Tuple[float, float, float]]

Sizes of all traffic signs.

min_distance: float

Minimal distance to generated sign.

max_distance: float

Maximal distance to generated sign.

sample_world_vec() → Vector

Sample vector to point in the fov.

_H(height: float) → matrix

Calculate transform from vehicle coords to img at a given height.

vehicle_point_to_img(vec: Vector) → ndarray

Transform vector in vehicle coords to pixel.

sign_coords(sign_name: str, angle_x: float, angle_z: float) → List[Vector]

Calculate rotated sign coordinates.

Parameters

• sign_name – Name of the sign.

• angle_x – Rotation around x axis.

• angle_z – Rotation around z axis.

kitcar_ml.utils.data_generation.utils module

Functions:

read_augmentation_data(...)	Read augmentation images and file names.
create_dataset(path_dataset_generated, ...)	Create dataset.
write_annotations_to_dataset(dataset, name, ...)

read_augmentation_data(path_to_augmentation_images: str) → Tuple[List[ndarray], List[str]]

Read augmentation images and file names.

create_dataset(path_dataset_generated: str, class_names: List[str]) → LabeledDataset

Create dataset.

Parameters

• path_dataset_generated – Path to dataset directory.

• class_names – Classes added to dataset.

write_annotations_to_dataset(dataset: LabeledDataset, name, sample_data)

Module contents