FastSurferCNN.data_loader.augmentation

class FastSurferCNN.data_loader.augmentation.AddGaussianNoise(mean=0, std=0.1)

Add gaussian noise to sample.

Attributes

std	Standard deviation.
mean	Gaussian mean.

Methods

__call__(sample)

Add gaussian noise to scalefactor.

class FastSurferCNN.data_loader.augmentation.AugmentationPadImage(pad_size=((16, 16), (16, 16)), pad_type='edge')

Pad Image with either zero padding or reflection padding of img, label and weight.

Attributes

pad_size_image	(tuple) The padding size for the image.
pad_size_mask	(tuple) The padding size for the mask.
pad_type	(str) The type of padding to be applied.

Methods

__call

Add zeroes.

class FastSurferCNN.data_loader.augmentation.AugmentationRandomCrop(output_size, crop_type='Random')

Randomly Crop Image to given size.

Methods

__call__(sample)

Crops the augmentation.

class FastSurferCNN.data_loader.augmentation.ToTensor

Convert ndarrays in sample to Tensors.

Methods

__call__(sample)

Convert the image to float within range [0, 1] and make it torch compatible.

class FastSurferCNN.data_loader.augmentation.ToTensorTest

Convert np.ndarrays in sample to Tensors.

Methods

__call__(img)

Convert the image to float within range [0, 1] and make it torch compatible.

class FastSurferCNN.data_loader.augmentation.ZeroPad2D(output_size, pos='top_left')

Pad the input with zeros to get output size.

Attributes

output_size	(Union[Number, Tuple[Number, Number]]) Size of the output image either as Number or tuple of two Number.
pos	(str, Optional) Position to put the input.

Methods

_pad(image)	Pad the input image with zeros.
__call__(sample)	Pad the image, label and weights.

class FastSurferCNN.data_loader.augmentation.ZeroPad2DTest(output_size, pos='top_left')

Pad the input with zeros to get output size.

Attributes

output_size	(Union[Number, Tuple[Number, Number]]) Size of the output image either as Number or tuple of two Number.
pos	(str) Position to put the input.

Methods

pad	Pad zeroes of image.
call	Call _pad().