Skip to content

Dataset

A dataset is an iterable that returns a sample for each tile by calling the data fetcher and the data preprocessor. The dataset is used by the data loader to fetch the samples for each batch.

Notes
  • A sample contains the data, the minimum x coordinate and the minimum y coordinate of a tile
  • The dataset is called concurrently by the data loader

Examples:

Assume the data fetcher, the coordinates and the data preprocessor are already created.

You can create a dataset and iterate over the samples.

>>> dataset = Dataset(
...     data_fetcher=data_fetcher,
...     coordinates=coordinates,
...     data_preprocessor=data_preprocessor,
... )
...
>>> for data, x_min, y_min in dataset:
...     ...
PARAMETER DESCRIPTION
data_fetcher

data fetcher

TYPE: DataFetcher

coordinates

coordinates (x_min, y_min) of each tile

TYPE: CoordinatesSet

data_preprocessor

data preprocessor

TYPE: DataPreprocessor | None DEFAULT: None