from typing import Any, Dict, Optional, Tuple, Union, cast
import torch
from torch import Tensor
from kornia.augmentation import random_generator as rg
from kornia.augmentation._2d.intensity.base import IntensityAugmentationBase2D
from kornia.geometry.bbox import bbox_generator, bbox_to_mask
[docs]class RandomErasing(IntensityAugmentationBase2D):
r"""Erase a random rectangle of a tensor image according to a probability p value.
.. image:: _static/img/RandomErasing.png
The operator removes image parts and fills them with zero values at a selected rectangle
for each of the images in the batch.
The rectangle will have an area equal to the original image area multiplied by a value uniformly
sampled between the range [scale[0], scale[1]) and an aspect ratio sampled
between [ratio[0], ratio[1])
Args:
scale: range of proportion of erased area against input image.
ratio: range of aspect ratio of erased area.
value: the value to fill the erased area.
same_on_batch: apply the same transformation across the batch.
p: probability that the random erasing operation will be performed.
keepdim: whether to keep the output shape the same as input (True) or broadcast it
to the batch form (False).
Shape:
- Input: :math:`(C, H, W)` or :math:`(B, C, H, W)`, Optional: :math:`(B, 3, 3)`
- Output: :math:`(B, C, H, W)`
Note:
Input tensor must be float and normalized into [0, 1] for the best differentiability support.
Additionally, this function accepts another transformation tensor (:math:`(B, 3, 3)`), then the
applied transformation will be merged int to the input transformation tensor and returned.
Examples:
>>> rng = torch.manual_seed(0)
>>> inputs = torch.ones(1, 1, 3, 3)
>>> aug = RandomErasing((.4, .8), (.3, 1/.3), p=0.5)
>>> aug(inputs)
tensor([[[[1., 0., 0.],
[1., 0., 0.],
[1., 0., 0.]]]])
To apply the exact augmenation again, you may take the advantage of the previous parameter state:
>>> input = torch.randn(1, 3, 32, 32)
>>> aug = RandomErasing((.4, .8), (.3, 1/.3), p=1.)
>>> (aug(input) == aug(input, params=aug._params)).all()
tensor(True)
"""
# Note: Extra params, inplace=False in Torchvision.
def __init__(
self,
scale: Union[Tensor, Tuple[float, float]] = (0.02, 0.33),
ratio: Union[Tensor, Tuple[float, float]] = (0.3, 3.3),
value: float = 0.0,
same_on_batch: bool = False,
p: float = 0.5,
keepdim: bool = False,
return_transform: Optional[bool] = None,
) -> None:
super().__init__(p=p, return_transform=return_transform, same_on_batch=same_on_batch, keepdim=keepdim)
self.scale = scale
self.ratio = ratio
self.value: float = float(value)
self._param_generator = cast(rg.RectangleEraseGenerator, rg.RectangleEraseGenerator(scale, ratio, float(value)))
def apply_transform(
self, input: Tensor, params: Dict[str, Tensor], flags: Dict[str, Any], transform: Optional[Tensor] = None
) -> Tensor:
_, c, h, w = input.size()
values = params["values"].unsqueeze(-1).unsqueeze(-1).unsqueeze(-1).repeat(1, *input.shape[1:]).to(input)
bboxes = bbox_generator(params["xs"], params["ys"], params["widths"], params["heights"])
mask = bbox_to_mask(bboxes, w, h) # Returns B, H, W
mask = mask.unsqueeze(1).repeat(1, c, 1, 1).to(input) # Transform to B, c, H, W
transformed = torch.where(mask == 1.0, values, input)
return transformed