prepare_dataset()

Prepare an .npz dataset from a directory of .npy cell image files.

Usage

Source

prepare_dataset(
    data_dir, output, channels=None, ignore=None, single_label=True
)

Parameters

data_dir: str

Directory containing .npy files and optionally metadata.json.

output: str

Full path for the output .npz file (e.g. ‘/path/to/rois.npz’).

channels: list[str] | None = None

Channel names. If None, read from metadata.json in data_dir.

ignore: list[str] | None = None

Labels to exclude (default: [‘unassigned’]).

single_label: bool = True

Skip masks with more than one label (default: True).