8.1.1.3.1.16. blueoil.datasets.open_images_v4

8.1.1.3.1.16.1. Module Contents

8.1.1.3.1.16.1.1. Classes

OpenImagesV4

Dataset base class

OpenImagesV4BoundingBox

Dataset base class

OpenImagesV4Classification

Dataset base class

OpenImagesV4BoundingBoxBase

Abstract class of dataset Open Images v4 format dataset.
class blueoil.datasets.open_images_v4.OpenImagesV4(*args, **kwargs)

Bases: blueoil.datasets.base.Base

Dataset base class

extend_dir = open_images_v4
available_subsets = ['train', 'validation', 'test']
task_extend =
property class_descriptions_csv(self)
property _classes_meta(self)
property classes(self)

Return the classes list in the data set.

property images_dir(self)
property num_per_epoch(self)

Returns the number of datas in the data subset.

property num_classes(self)

Return the number of classes in the data set.

property files_and_annotations(self)
class blueoil.datasets.open_images_v4.OpenImagesV4BoundingBox(class_level=0, *args, **kwargs)

Bases: blueoil.datasets.open_images_v4.OpenImagesV4, blueoil.datasets.base.ObjectDetectionBase

Dataset base class

task_extend = bounding_boxes
available_subsets = ['train', 'validation', 'test']
property annotations_csv(self)
property classes(self)

Return the classes list in the data set.

property _target_labels(self)

Map of {csv raw label name: Be mapped target label name}.

_make_target_labels(self)
_search_subcategory(self, parent_label_name, d, level)
_bboxes(self)
property files_and_annotations(self)
_files_and_annotations(self)
classmethod count_max_boxes(cls)

Count max boxes size over all subsets.

property num_max_boxes(self)

Return count max box size of available subsets.

__getitem__(self, i)

Returns the i-th item of the dataset.

__len__(self)

returns the number of items in the dataset.

class blueoil.datasets.open_images_v4.OpenImagesV4Classification(batch_size=100, *args, **kwargs)

Bases: blueoil.datasets.open_images_v4.OpenImagesV4

Dataset base class

task_extend = classification
available_subsets = ['train', 'validation', 'test']
property annotations_csv(self)
property files_and_annotations(self)
__getitem__(self, i)

Returns the i-th item of the dataset.

__len__(self)

returns the number of items in the dataset.

class blueoil.datasets.open_images_v4.OpenImagesV4BoundingBoxBase(*args, **kwargs)

Bases: blueoil.datasets.base.StoragePathCustomizable, blueoil.datasets.open_images_v4.OpenImagesV4BoundingBox

Abstract class of dataset Open Images v4 format dataset.

structure like

$DATA_DIR/extend_dir/class-descriptions.csv $DATA_DIR/extend_dir/annotations-bbox.csv $DATA_DIR/extend_dir/images/xxxa.jpeg $DATA_DIR/extend_dir/images/yyyb.png $DATA_DIR/extend_dir/images/123.jpg $DATA_DIR/extend_dir/images/023.jpg $DATA_DIR/extend_dir/images/wwww.jpg

When child class has validation_extend_dir, the validation subset consists from the folders.

$DATA_DIR/validation_extend_dir/annotations-bbox.csv $DATA_DIR/validation_extend_dir/images/xxxa.jpeg $DATA_DIR/validation_extend_dir/images/yyyb.png $DATA_DIR/validation_extend_dir/images/123.jpg $DATA_DIR/validation_extend_dir/images/023.jpg $DATA_DIR/validation_extend_dir/images/wwww.jpg

task_extend =
property class_descriptions_csv(self)
property annotations_csv(self)
property images_dir(self)
property classes(self)

Return the classes list in the data set.

property files_and_annotations(self)
_files_and_annotations(self)