8.1.1.3.1.22. blueoil.datasets.tfds

8.1.1.3.1.22.1. Module Contents

8.1.1.3.1.22.1.1. Classes

TFDSMixin

A Mixin to compose dataset classes for TFDS.

TFDSClassification

A dataset class for loading TensorFlow Datasets for classification.

TFDSObjectDetection

A dataset class for loading TensorFlow Datasets for object detection.

TFDSSegmentation

A dataset class for loading TensorFlow Datasets for segmentation.

8.1.1.3.1.22.1.2. Functions

_grayscale_to_rgb(record)

_format_classification_record(record, image_size, num_classes)

_format_object_detection_record(record, image_size, num_max_boxes)

_format_segmentation_record(record, image_size)

blueoil.datasets.tfds._grayscale_to_rgb(record)
blueoil.datasets.tfds._format_classification_record(record, image_size, num_classes)
blueoil.datasets.tfds._format_object_detection_record(record, image_size, num_max_boxes)
blueoil.datasets.tfds._format_segmentation_record(record, image_size)
class blueoil.datasets.tfds.TFDSMixin(name, data_dir, image_size, download=False, local_rank=None, num_max_boxes=None, tfds_pre_processor=None, tfds_augmentor=None, *args, **kwargs)

A Mixin to compose dataset classes for TFDS.

available_subsets = ['train', 'validation']
extend_dir
property num_per_epoch(self)
property __getitem__(self, i)
__len__(self)
_init_available_splits(self)

Initializing available splits dictionary depending on what kind of splits the dataset has.

abstract _validate_feature_structure(self)

Checking if the given dataset has a valid feature structure.

This method will raise a ValueError if the structure is invalid.

Args:

Returns:

abstract _format_dataset(self)

Converting the format of loaded dataset.

class blueoil.datasets.tfds.TFDSClassification(name, data_dir, image_size, download=False, local_rank=None, num_max_boxes=None, tfds_pre_processor=None, tfds_augmentor=None, *args, **kwargs)

Bases: blueoil.datasets.tfds.TFDSMixin, blueoil.datasets.base.Base

A dataset class for loading TensorFlow Datasets for classification. TensorFlow Datasets which have “label” and “image” features can be loaded by this class.

builder_class
property classes(self)

Return the classes list in the data set.

property num_classes(self)

Return the number of classes in the data set.

_validate_feature_structure(self)

Checking if the given dataset has a valid feature structure.

This method will raise a ValueError if the structure is invalid.

Args:

Returns:

_format_dataset(self)

Converting the format of loaded dataset.

class blueoil.datasets.tfds.TFDSObjectDetection(name, data_dir, image_size, download=False, local_rank=None, num_max_boxes=None, tfds_pre_processor=None, tfds_augmentor=None, *args, **kwargs)

Bases: blueoil.datasets.tfds.TFDSMixin, blueoil.datasets.base.ObjectDetectionBase

A dataset class for loading TensorFlow Datasets for object detection. TensorFlow Datasets which have “objects” and “image” features can be loaded by this class.

builder_class
classmethod count_max_boxes(cls, builder, local_rank)

Count max boxes size over all subsets.

property classes(self)

Return the classes list in the data set.

property num_classes(self)

Return the number of classes in the data set.

property num_max_boxes(self)

Return count max box size of available subsets.

_validate_feature_structure(self)

Checking if the given dataset has a valid feature structure.

This method will raise a ValueError if the structure is invalid.

Args:

Returns:

_format_dataset(self)

Converting the format of loaded dataset.

class blueoil.datasets.tfds.TFDSSegmentation(name, data_dir, image_size, download=False, local_rank=None, num_max_boxes=None, tfds_pre_processor=None, tfds_augmentor=None, *args, **kwargs)

Bases: blueoil.datasets.tfds.TFDSMixin, blueoil.datasets.base.SegmentationBase

A dataset class for loading TensorFlow Datasets for segmentation. TensorFlow Datasets which have “label” and “image” features can be loaded by this class.

builder_class
property classes(self)

Return the classes list in the data set.

property num_classes(self)

Return the number of classes in the data set.

_validate_feature_structure(self)

Checking if the given dataset has a valid feature structure.

This method will raise a ValueError if the structure is invalid.

Args:

Returns:

_format_dataset(self)

Converting the format of loaded dataset.