fiftyone.core.map.mapper#

Abstract mapping backend

Copyright 2017-2025, Voxel51, Inc.
voxel51.com

Functions:

check_if_return_is_sample(sample_collection, ...)

Check if the map function returns a sample

Classes:

`Mapper`(batch_cls, num_workers[, batch_size, ...])	Base class for mapping samples in parallel
`LocalMapper`(batch_cls, num_workers[, ...])	Base class for mapping samples in parallelizing on the same machine

fiftyone.core.map.mapper.check_if_return_is_sample(sample_collection: SampleCollection[T], map_fcn: Callable[[T], R]) → bool#: Check if the map function returns a sample

class fiftyone.core.map.mapper.Mapper(batch_cls: Type[SampleBatch], num_workers: int, batch_size: int | None = None, default_progress: bool | None = None)#

Bases: ABC

Base class for mapping samples in parallel

Methods:

`create`(*, config, batch_cls[, num_workers, ...])	Create a new mapper instance
`map_samples`(sample_collection, map_fcn, *[, ...])	Applies map function to each sample and returns an iterator of the results.

Attributes:

`num_workers`	Number of workers to use
`batch_size`	Number of samples per worker batch

classmethod create(*, config: FiftyOneConfig, batch_cls: Type[SampleBatch], num_workers: int | None = None, batch_size: int | None = None, **__)#: Create a new mapper instance

property num_workers: int#: Number of workers to use

property batch_size: int | None#: Number of samples per worker batch

map_samples(sample_collection: SampleCollection[T], map_fcn: Callable[[T | U], R], *, iter_fcn: Callable[[SampleCollection[T]], Iterable[Tuple[ObjectId, U]]] | None = None, progress: bool | Literal['workers'] | None = None, save: bool = False, skip_failures: bool = True) → Iterator[Tuple[ObjectId, R]]#

Applies map function to each sample and returns an iterator of the results.

Parameters:

sample_collection (SampleCollection[T]) – The sample collection to map.
map_fcn (Callable[[Union[T, U]], R]) – The map function to apply to each sample.
iter_fcn (Callable[[T], U]) – The function to iterate over the sample collection. If not provided the iteration function used is iter_samples.
progress (Union[bool, Literal["workers"]]) – Whether or not and how to render progress.
save (bool, optional) – Whether to save mutated samples mutated in the map function. Only valid when using the iter_fcn is None. Defaults to False.
skip_failures (bool, optional) – Whether to gracefully continue without raising an error if the map function raises an exception for a sample. Defaults to True.

Yields:

Iterator[Tuple[bson.ObjectId, R]] –

The sample ID and the result of: the map function for the sample.

class fiftyone.core.map.mapper.LocalMapper(batch_cls: Type[SampleBatch], num_workers: int, batch_size: int | None = None, default_progress: bool | None = None)#

Bases: Mapper, ABC

Base class for mapping samples in parallelizing on the same machine