Image-Level Transformations

GridShuffle

class augmax.GridShuffle(grid_size: Tuple[int, int] | int = (4, 4), p: float = 0.5, input_types=InputType.IMAGE)[source]

Divides the image into grid cells and shuffles them randomly.

Parameters:

grid_size (int, int) – Tuple of (gridcells_x, gridcells_y) that specifies into how many cells the image is to be divided along each axis. If only a single number is given, that value will be used along both axes. Currently requires that each image dimension is a multiple of the corresponding value.
p (float) – Probability of applying the transformation

../_images/GridShuffle_0.png — Augmentation Examples for GridShuffle(p=1.0)

../_images/GridShuffle_1.png — Augmentation Examples for GridShuffle(p=1.0)

Blur

class augmax.Blur(size: int = 5, p: float = 0.5)[source]

../_images/Blur_0.png — Augmentation Examples for Blur(p=1.0)

../_images/Blur_1.png — Augmentation Examples for Blur(p=1.0)

GaussianBlur

class augmax.GaussianBlur(sigma: int = 3, p: float = 0.5)[source]

../_images/GaussianBlur_0.png — Augmentation Examples for GaussianBlur(p=1.0)

../_images/GaussianBlur_1.png — Augmentation Examples for GaussianBlur(p=1.0)