def load_data(
self,
paths: Union[str, List[str]],
targets: Optional[List[str]] = None,
sampling_rate: int = 16000,
) -> Sequence:
self.sampling_rate = sampling_rate
if targets is None:
return to_samples(list_valid_files(paths, AUDIO_EXTENSIONS))
return to_samples(*filter_valid_files(
paths, targets, valid_extensions=AUDIO_EXTENSIONS))
def load_data(
self,
array: Any,
masks: Any = None,
num_classes: Optional[int] = None,
labels_map: Optional[Dict[int, Tuple[int, int, int]]] = None,
) -> List[Dict[str, Any]]:
self.load_labels_map(num_classes, labels_map)
return to_samples(array, masks)
def load_data(
self,
tensor: Any,
targets: Optional[List[Any]] = None,
target_formatter: Optional[TargetFormatter] = None
) -> List[Dict[str, Any]]:
if targets is not None:
self.load_target_metadata(targets,
target_formatter=target_formatter)
return to_samples(tensor, targets)
def load_data(
self,
files: List[PATH_TYPE],
targets: Optional[List[Any]] = None,
sampling_rate: int = 16000,
n_fft: int = 400,
target_formatter: Optional[TargetFormatter] = None,
) -> List[Dict[str, Any]]:
self.sampling_rate = sampling_rate
self.n_fft = n_fft
if targets is None:
files = filter_valid_files(files,
valid_extensions=AUDIO_EXTENSIONS +
IMG_EXTENSIONS + NP_EXTENSIONS)
return to_samples(files)
files, targets = filter_valid_files(files,
targets,
valid_extensions=AUDIO_EXTENSIONS +
IMG_EXTENSIONS + NP_EXTENSIONS)
self.load_target_metadata(targets, target_formatter=target_formatter)
return to_samples(files, targets)
def load_data(
self,
files: Union[PATH_TYPE, List[PATH_TYPE]],
mask_files: Optional[Union[PATH_TYPE, List[PATH_TYPE]]] = None,
num_classes: Optional[int] = None,
labels_map: Optional[Dict[int, Tuple[int, int, int]]] = None,
) -> List[Dict[str, Any]]:
self.load_labels_map(num_classes, labels_map)
if mask_files is None:
files = filter_valid_files(files, valid_extensions=IMG_EXTENSIONS)
else:
files, mask_files = filter_valid_files(
files, mask_files, valid_extensions=IMG_EXTENSIONS)
return to_samples(files, mask_files)
def load_data(
self,
files: List[PATH_TYPE],
targets: Optional[List[Any]] = None,
target_formatter: Optional[TargetFormatter] = None,
) -> List[Dict[str, Any]]:
if targets is None:
return super().load_data(files)
files, targets = filter_valid_files(files,
targets,
valid_extensions=IMG_EXTENSIONS +
NP_EXTENSIONS)
self.load_target_metadata(targets, target_formatter=target_formatter)
return to_samples(files, targets)
def load_data(
self,
sample_collection: SampleCollection,
label_field: str = "ground_truth",
num_classes: Optional[int] = None,
labels_map: Optional[Dict[int, Tuple[int, int, int]]] = None,
) -> List[Dict[str, Any]]:
self.load_labels_map(num_classes, labels_map)
self.label_field = label_field
label_utilities = FiftyOneLabelUtilities(label_field, fo.Segmentation)
label_utilities.validate(sample_collection)
self._fo_dataset_name = sample_collection.name
return to_samples(sample_collection.values("filepath"))
def load_data(
self,
examples: Collection[np.ndarray],
targets: Optional[Sequence[Any]] = None,
target_formatter: Optional[TargetFormatter] = None,
) -> Sequence[Dict[str, Any]]:
"""Sets the ``num_features`` attribute and calls ``super().load_data``.
Args:
examples: The ``np.ndarray`` (num_examples x num_features).
targets: Associated targets.
target_formatter: Optionally provide a ``TargetFormatter`` to control how targets are formatted.
Returns:
A sequence of samples / sample metadata.
"""
if not self.predicting and isinstance(examples, np.ndarray):
self.num_features = examples.shape[1]
if targets is not None:
self.load_target_metadata(targets,
target_formatter=target_formatter)
return to_samples(examples, targets)
def predict_load_data(
self,
sample_collection: SampleCollection,
) -> List[Dict[str, Any]]:
return to_samples(sample_collection.values("filepath"))
def load_data(self, tensor: Any) -> List[Dict[str, Any]]:
return to_samples(tensor)
def load_data(self, files: List[PATH_TYPE]) -> List[Dict[str, Any]]:
files = filter_valid_files(files,
valid_extensions=IMG_EXTENSIONS +
NP_EXTENSIONS)
return to_samples(files)
def load_data(self, array: Any) -> List[Dict[str, Any]]:
return to_samples(array)
