AlbuStrongAugment
Bases: BaseCompose
__init__ ¶
__init__(augment_space: Dict[str, tuple] = AUGMENT_SPACE, operations: Tuple[int] = (3, 4, 5), probabilites: Tuple[float] = (0.2, 0.3, 0.5), seed: Optional[int] = None, p=1.0) -> None
Strong augment augmentation policy albumentations wrapper.
Augment like there's no tomorrow: Consistently performing neural networks for medical imaging: https://arxiv.org/abs/2206.15274
Parameters¶
augment_space : Dict[str, tuple], default: AUGMENT_SPACE
Augmentation space to sample operations from.
operations : Tuple[int], default: [3, 4, 5].
Number of operations to apply. If None, sample from
[1, len(augment_space)].
probabilites : Tuple[float], default: [0.2, 0.3, 0.5]
Probabilities of sampling operations. If None, sample from
the uniform distribution.
seed : Optional[int], default: None
Random seed.
p : float, default: 1.0
Probability of applying the transform.
__call__ ¶
Apply the StrongAugment transformation pipeline.