scallops.io.to_label_crops

scallops.io.to_label_crops(intensity_image, label_image, df, output_dir, crop_size=(224, 224), output_format='tiff', centroid_cols=None, label_col=None, gaussian_sigma=None)

Export individual label crops as tiff or npy files.

Parameters:

  • intensity_image (Array | Array) – Image data

  • label_image (Array | Array) – Label data. If supplied, image is masked by corresponding segmentation mask

  • df (DataFrame) – DataFrame containing centroid_cols and label_col. Typically output from find-objects.

  • crop_size (tuple[int, int]) – Size of label crop

  • output_dir (str) – Crop output directory

  • output_format (Literal['tiff', 'npy']) – Crop output format

  • centroid_cols (Sequence[str] | None) – Columns in df containing y and x centroids.

  • label_col (str | None) – Column in df containing label id. Set to None if label is found in data frame index.

  • gaussian_sigma (float | None) – If not None, apply gaussian-smoothed mask to isolate target mask

Returns:

Input data frame with objects at well edges removed. Adds columns crop_url, the URL to

Return type:

DataFrame

output label crops, and crop-bbox-0, crop-bbox-2, crop-bbox-1, crop-bbox-3 containing the coordinates of the crop.