fiftyone.utils.fiw

module documentation

(source)

Utilities for working with the Families in the Wild dataset.

Class	`FIWDatasetImporter`	Importer for Faces in the Wild-formatted datasets stored on disk.
Function	`download_fiw_dataset`	Downloads and extracts the Families in the Wild dataset.
Function	`get_identifier_filepaths_map`	Creates a mapping of `family_id/member_id` identifier to a list of filepaths for each person.
Function	`get_pairwise_labels`	Gets a list of all pairs of people that are related and the label of their relation, either through the "kinships" or "relationships" field.
Variable	`logger`	Undocumented
Function	`_download_images_if_necessary`	Undocumented
Function	`_download_labels_if_necessary`	Undocumented
Function	`_format_mid`	Undocumented
Function	`_get_dataset_info`	Undocumented
Function	`_get_gender`	Undocumented
Function	`_get_mid`	Undocumented
Function	`_get_name`	Undocumented
Function	`_get_picture_face`	Undocumented
Function	`_get_relationships`	Undocumented
Function	`_is_missing_images`	Undocumented
Function	`_is_missing_labels`	Undocumented
Function	`_load_member`	Undocumented
Function	`_load_split`	Undocumented
Function	`_organize_data`	Undocumented
Function	`_organize_labels`	Undocumented
Function	`_parse_identifier`	Undocumented
Function	`_parse_kinship_map`	Undocumented
Constant	`_IMAGES_DOWNLOAD_LINK`	Undocumented
Constant	`_LISTS_DOWNLOAD_LINK`	Undocumented
Constant	`_MD5_DATA_DOWNLOAD_LINK`	Undocumented
Constant	`_MD5_LISTS_DOWNLOAD_LINK`	Undocumented
Constant	`_RELATIONSHIP_MAP`	Undocumented
Constant	`_SPLITS`	Undocumented

def download_fiw_dataset(dataset_dir, split, scratch_dir=None, cleanup=False): (source) ¶

Downloads and extracts the Families in the Wild dataset.

Any existing files are not re-downloaded.

Parameters
dataset_dir	the directory to output the final dataset
split	the split being loaded
scratch_dir:`None`	a scratch directory to use to store temporary files
cleanup:`True`	whether to cleanup the scratch directory after extraction

def get_identifier_filepaths_map(samples): (source) ¶

Creates a mapping of family_id/member_id identifier to a list of filepaths for each person.

Example:

{
    "F0325/MID4": [
        "/path/to/fiftyone/fiw/train/data/F0325/MID4/P03451_face4.jpg",
        ...
    ],
    ...
}

Parameters
samples	a `fiftyone.core.collections.SampleCollection`
Returns
a dict mapping `family_id/member_id` identifiers to a list of filepaths containing images of the corresponding person

def get_pairwise_labels(samples, label_type='kinships'): (source) ¶

Gets a list of all pairs of people that are related and the label of their relation, either through the "kinships" or "relationships" field.

Example:

[
   ["F0009/MID2", "F0009/MID4", "sibling"],
   ...
]

Parameters
samples	a `fiftyone.core.collections.SampleCollection`
label_type:"kinships"	the type of label of which to return pairwise listings options are `("kinships", "relationships")`
Returns
a list of triplets containing the identifier of person 1, identifier of person 2, and their kinship or relationship

logger = (source) ¶

Undocumented

def _download_images_if_necessary(dataset_dir, scratch_dir): (source) ¶

Undocumented

def _download_labels_if_necessary(dataset_dir, scratch_dir): (source) ¶

Undocumented

def _format_mid(member_id_int): (source) ¶

Undocumented

def _get_dataset_info(dataset_dir, split): (source) ¶

Undocumented

def _get_gender(labels, member_id): (source) ¶

Undocumented

def _get_mid(member_id): (source) ¶

Undocumented

def _get_name(labels, member_id): (source) ¶

Undocumented

def _get_picture_face(img_path): (source) ¶

Undocumented

def _get_relationships(labels, member_id_int, num_members, family_id): (source) ¶

Undocumented

def _is_missing_images(dataset_dir): (source) ¶

Undocumented

def _is_missing_labels(dataset_dir): (source) ¶

Undocumented

def _load_member(dir_images, family_id, member_id, labels, face_indices, tags, kinship_map): (source) ¶

Undocumented

def _load_split(split_dir, tags, labels_path): (source) ¶

Undocumented

def _organize_data(unzip_path, dataset_dir): (source) ¶

Undocumented

def _organize_labels(scratch_dir, dataset_dir): (source) ¶

Undocumented

def _parse_identifier(identifier): (source) ¶

Undocumented

def _parse_kinship_map(labels_path): (source) ¶

Undocumented

_IMAGES_DOWNLOAD_LINK: str = (source) ¶

Undocumented

Value

'1rkrDGOjS0e_pptzRHZl5bRGq0yy5xQxQ'

_LISTS_DOWNLOAD_LINK: str = (source) ¶

Undocumented

Value

'1nt22yiCfdGF7aIguUb-SJmsM_1CvcYjg'

_MD5_DATA_DOWNLOAD_LINK: str = (source) ¶

Undocumented

Value

'121lbbeaiY-qM2tK9sJXWNuvMczVuwi2p'

_MD5_LISTS_DOWNLOAD_LINK: str = (source) ¶

Undocumented

Value

'1nt22yiCfdGF7aIguUb-SJmsM_1CvcYjg'

_RELATIONSHIP_MAP: dict[int, str] = (source) ¶

Undocumented

Value

{1: 'child',
 2: 'sibling',
 3: 'grandchild',
 4: 'parent',
 5: 'spouse',
 6: 'grandparent',
 7: 'great grandchild',
...

_SPLITS: list[str] = (source) ¶

Undocumented

Value

['train', 'test', 'val']