Skip to content

Download Datasets

The download command is used to download all datasets specified. Please refer to Datasets for details on the available datasets. Additional datasets can be added by creating a new dataset class and registering it with HeartKit dataset factory.

Usage

Example

The following command will download and prepare four datasets.

heartkit -m download -c ./configs/download-datasets.json
# ^ No task is required

from pathlib import Path
import heartkit as hk

hk.datasets.download_datasets(hk.HKDownloadParams(
    ds_path=Path("./datasets"),
    datasets=["icentia11k", "ludb", "qtdb", "synthetic"],
    progress=True
))

Arguments

The following table lists the arguments that can be used with the download command. All datasets will be saved in their own subdirectory within the ds_path directory.

Argument Type Opt/Req Default Description
ds_path Path Optional Path() Dataset root directory
datasets list[DatasetTypes] Optional Datasets
progress bool Optional True Display progress bar
force bool Optional False Force download dataset- overriding existing files
data_parallelism int Optional lambda: os.cpu_count() or 1 # of data loaders running in parallel