neodroidvision.data.detection.voc.voc.VOCDataset

class neodroidvision.data.detection.voc.voc.VOCDataset(data_root: Path, dataset_name: str, split: SplitEnum, img_transform: Optional[callable] = None, annotation_transform: Optional[callable] = None)[source]

Bases: ObjectDetectionDataset

description

__init__(data_root: Path, dataset_name: str, split: SplitEnum, img_transform: Optional[callable] = None, annotation_transform: Optional[callable] = None)[source]

Dataset for VOC data.

data_root: the root of the VOC2007 or VOC2012 dataset, the directory contains the following

Annotations, ImageSets, JPEGImages, SegmentationClass, SegmentationObject.

Parameters

  • data_root –

  • dataset_name –

  • split –

  • img_transform –

  • annotation_transform –

  • keep_difficult –

Methods

__init__(data_root, dataset_name, split[, ...])

Dataset for VOC data.

get_annotation(index)

param index

get_img_info(index)

param index

Attributes

categories

data_dirs

predictor_shape

description

response_shape

param self

split_names

return: :rtype:

splits

get_annotation(index)[source]
Parameters

index –

Returns

Return type

get_img_info(index)[source]
Parameters

index –

Returns

Return type

property predictor_shape: Tuple[int, ...]

description

property response_shape: Tuple[int, ...]
Parameters

self –

Returns:

property split_names: Dict[SplitEnum, str]

return: :rtype: