yamle.methods.augmentation_classification module#

class yamle.methods.augmentation_classification.AugmentationImageClassificationMethod(*args, **kwargs)[source]#

Bases: BaseMethod

This is the base class for image classification methods with augmentations.

tasks = ['classification']#
class yamle.methods.augmentation_classification.CutOutImageClassificationMethod(batch_proportion=0.5, cutout_size=16, *args, **kwargs)[source]#

Bases: AugmentationImageClassificationMethod

This is the base class for image classification methods with cutout.

Parameters:

  • batch_proportion (float) – The proportion of the batch to cut out.

  • cutout_size (int) – The size of the cutout.

static add_specific_args(parent_parser)[source]#

Add arguments for CutOutImageClassificationMethod.

Return type:

ArgumentParser

test_name: Optional[str]#
prepare_data_per_node: bool#
allow_zero_length_dataloader_with_multiple_devices: bool#
training: bool#
class yamle.methods.augmentation_classification.MixUpImageClassificationMethod(batch_proportion=0.5, mixup_alpha=0.4, *args, **kwargs)[source]#

Bases: AugmentationImageClassificationMethod

This is the base class for image classification methods with mixup.

Parameters:

  • batch_proportion (float) – The proportion of the batch to mixup.

  • mixup_alpha (float) – The alpha parameter for the beta distribution.

static add_specific_args(parent_parser)[source]#

Add arguments for MixUpImageClassificationMethod.

Return type:

ArgumentParser

test_name: Optional[str]#
prepare_data_per_node: bool#
allow_zero_length_dataloader_with_multiple_devices: bool#
training: bool#
class yamle.methods.augmentation_classification.CutMixImageClassificationMethod(batch_proportion=0.5, cutmix_alpha=1.0, *args, **kwargs)[source]#

Bases: AugmentationImageClassificationMethod

This is the base class for image classification methods with cutmix.

Parameters:

  • batch_proportion (float) – The proportion of the batch to cutmix.

  • cutmix_alpha (float) – The alpha parameter for the beta distribution.

static add_specific_args(parent_parser)[source]#

Add arguments for CutMixImageClassificationMethod.

Return type:

ArgumentParser

test_name: Optional[str]#
prepare_data_per_node: bool#
allow_zero_length_dataloader_with_multiple_devices: bool#
training: bool#
class yamle.methods.augmentation_classification.RandomErasingImageClassificationMethod(batch_proportion=0.5, random_erasing_scale=(0.02, 0.33), random_erasing_ratio=(0.3, 3.3), random_erasing_value=0.0, *args, **kwargs)[source]#

Bases: AugmentationImageClassificationMethod

This is the base class for image classification methods with random erasing.

Parameters:

  • batch_proportion (float) – The proportion of the batch to randomly erase.

  • random_erasing_scale (Tuple[float, float]) – The range of the random erasing scale.

  • random_erasing_ratio (Tuple[float, float]) – The range of the random erasing ratio.

  • random_erasing_value (float) – The value to fill the erased area with.

static add_specific_args(parent_parser)[source]#

Add arguments for RandomErasingImageClassificationMethod.

Return type:

ArgumentParser

test_name: Optional[str]#
prepare_data_per_node: bool#
allow_zero_length_dataloader_with_multiple_devices: bool#
training: bool#