cellmap_data.transforms.augment.normalize

Classes

Normalize([shift, scale])

Initialize the normalization transformation.

class cellmap_data.transforms.augment.normalize.Normalize(shift=0, scale=0.00392156862745098)

Initialize the normalization transformation. :param shift: Shift values, before scaling. Defaults to 0. :type shift: float, optional :param scale: Scale values after shifting. Defaults to 1/255. :type scale: float, optional

This is helpful in normalizing the input to the range [0, 1], especially for data saved as uint8 which is scaled to [0, 255].

Example

>>> import torch
>>> from cellmap_data.transforms.augment import Normalize
>>> x = torch.tensor([[0, 255], [2, 3]], dtype=torch.uint8)
>>> Normalize(shift=0, scale=1/255).transform(x, {})
tensor([[0.0000, 1],
        [0.0078, 0.0118]])

transform(x: Any, params: Dict[str, Any] | None = None) → Any

Transform the input.

Parameters:

• x (Any)

• params (Dict[str, Any] | None)

Return type:

Any