mednet.models.transforms

Functions

crop_image_to_mask(img, mask)	Square crop image to the boundaries of a boolean mask.
square_center_pad(img, size)	Return a squared version of the image, centered on a canvas padded with zeros.

Classes

SquareCenterPad()

Transform to a squared version of the image, centered on a canvas padded with zeros.

mednet.models.transforms.crop_image_to_mask(img, mask)

Square crop image to the boundaries of a boolean mask.

Parameters:

• img (Tensor) – The image to crop, of shape channels x height x width.

• mask (Tensor) – The boolean mask to use for cropping.

Return type:

Tensor

Returns:

The cropped image.

mednet.models.transforms.square_center_pad(img, size)

Return a squared version of the image, centered on a canvas padded with zeros.

Parameters:

• img (Tensor) – The tensor to be transformed. Expected to be in the form: [..., [1,3], H, W] (i.e. arbitrary number of leading dimensions).

• size (Any) – Height and width of the image.

Return type:

Tensor

Returns:

Transformed tensor, guaranteed to be square (ie. equal height and width).

class mednet.models.transforms.SquareCenterPad

Bases: Transform

Transform to a squared version of the image, centered on a canvas padded with zeros.