steps.base.dataset.loader

loader

Functions:

Name	Description
load_coco_datasets	A step for loading COCO datasets based on the current pipeline context (training or processing).
load_yolo_datasets	A step for loading YOLO datasets based on the current pipeline context (training or processing).

load_coco_datasets(use_id=True, skip_asset_listing=False)

A step for loading COCO datasets based on the current pipeline context (training or processing).

This function adapts to different contexts and loads datasets accordingly: - Training Contexts: Loads datasets for training, validation, and testing splits. - Processing Contexts: Loads either a single dataset or multiple datasets depending on the context.

Parameters:

Name	Type	Description	Default
use_id	bool	Whether to use asset UUIDs as filenames when downloading images/annotations. If False, filenames from the original dataset will be used. Default is True.	True
skip_asset_listing	bool	Whether to skip listing dataset assets before downloading. Default is False. This is applicable only for processing contexts.	False

Returns:

Type

Description

DatasetCollection[CocoDataset] | CocoDataset

Union[DatasetCollection[CocoDataset], CocoDataset]: The loaded dataset(s) based on the context. For Training Contexts: Returns a DatasetCollection[CocoDataset] containing training, validation, and test datasets. For Processing Contexts: If both input and output datasets are available, returns a DatasetCollection[CocoDataset]. If only an input dataset is available, returns a single CocoDataset for the input dataset.

Example

• In a Training Context, the function loads and prepares datasets for training, validation, and testing.
• In a Processing Context, it loads the input and output datasets (if available) or just the input dataset.

load_yolo_datasets(use_id=True, skip_asset_listing=False)

A step for loading YOLO datasets based on the current pipeline context (training or processing).

This function adapts to different contexts and loads datasets accordingly: - Training Contexts: Loads datasets for training, validation, and testing splits. - Processing Contexts: Loads either a single dataset or multiple datasets depending on the context.

Parameters:

Name	Type	Description	Default
use_id	bool	Whether to use asset UUIDs as filenames when downloading images/annotations. If False, filenames from the original dataset will be used. Default is True.	True
skip_asset_listing	bool	Whether to skip listing dataset assets before downloading. Default is False. This is applicable only for processing contexts.	False

Returns:

Type	Description
DatasetCollection[YoloDataset] | YoloDataset	Union[DatasetCollection[YoloDataset], YoloDataset]: The loaded dataset(s) based on the context.