fishy.data.augmentation

Data augmentation module for the deep learning pipeline.

Classes

class fishy.data.augmentation.AugmentationConfig(enabled: bool = False, num_augmentations: int = 5, noise_enabled: bool = True, shift_enabled: bool = False, scale_enabled: bool = False, crop_enabled: bool = False, flip_enabled: bool = False, permutation_enabled: bool = False, noise_level: float = 0.1, shift_range: float = 0.1, scale_range: float = 0.1, crop_size: float = 0.8)

Bases: object

Configuration for data augmentation.

Examples

>>> config = AugmentationConfig(enabled=True, num_augmentations=10)
>>> config.enabled
True
>>> config.num_augmentations
10

__init__(enabled: bool = False, num_augmentations: int = 5, noise_enabled: bool = True, shift_enabled: bool = False, scale_enabled: bool = False, crop_enabled: bool = False, flip_enabled: bool = False, permutation_enabled: bool = False, noise_level: float = 0.1, shift_range: float = 0.1, scale_range: float = 0.1, crop_size: float = 0.8) → None

crop_enabled: bool = False

crop_size: float = 0.8

enabled: bool = False

flip_enabled: bool = False

noise_enabled: bool = True

noise_level: float = 0.1

num_augmentations: int = 5

permutation_enabled: bool = False

scale_enabled: bool = False

scale_range: float = 0.1

shift_enabled: bool = False

shift_range: float = 0.1

class fishy.data.augmentation.DataAugmenter(config: AugmentationConfig)

Bases: object

Applies various data augmentation techniques to a DataLoader.

__init__(config: AugmentationConfig) → None

augment(dataloader: DataLoader) → DataLoader

s