Erasing

Targets:

image

mask

bboxes

keypoints

Image Types:uint8, float32

Randomly erase rectangular regions (Random Erasing). area_ratio_range, aspect_ratio_range, fill. Improves robustness; common in image classification.

This augmentation helps improve model robustness by randomly masking out rectangular regions in the image, simulating occlusions and encouraging the model to learn from partial information. It's particularly effective for image classification and person re-identification tasks.

Arguments

scale

tuple[float, float]

[0.02,0.33]

Range for the proportion of image area to erase. The actual area will be randomly sampled from (scale[0] * image_area, scale[1] * image_area). Default: (0.02, 0.33)

ratio

tuple[float, float]

[0.3,3.3]

Range for the aspect ratio (width/height) of the erased region. The actual ratio will be randomly sampled from (ratio[0], ratio[1]). Default: (0.3, 3.3)

fill

Value used to fill the erased regions. Can be:

int or float: fills all channels with this value
tuple: fills each channel with corresponding value
"random": fills each pixel with random values
"random_uniform": fills entire erased region with a single random color
"inpaint_telea": uses OpenCV Telea inpainting method
"inpaint_ns": uses OpenCV Navier-Stokes inpainting method Default: 0

fill_mask

tuple[float, ...] | float | None

Value used to fill erased regions in the mask. If None, mask regions are not modified. Default: None

p

float

0.5

Probability of applying the transform. Default: 0.5

Examples

>>> import numpy as np
>>> import albumentations as A
>>> image = np.random.randint(0, 256, (100, 100, 3), dtype=np.uint8)
>>> # Basic usage with default parameters
>>> transform = A.Erasing()
>>> transformed = transform(image=image)
>>> # Custom configuration
>>> transform = A.Erasing(
...     scale=(0.1, 0.4),
...     ratio=(0.5, 2.0),
...     fill_value="random_uniform",
...     p=1.0
... )
>>> transformed = transform(image=image)

Notes

The transform attempts to find valid erasing parameters up to 10 times. If unsuccessful, no erasing is performed.
The actual erased area and aspect ratio are randomly sampled within the specified ranges for each application.
When using inpainting methods, only grayscale or RGB images are supported.

References

Paperhttps://arxiv.org/abs/1708.04896
Implementation inspired by torchvisionhttps://pytorch.org/vision/stable/transforms.html#torchvision.transforms.RandomErasing

>>> import numpy as np >>> import albumentations as A >>> image = np.random.randint(0, 256, (100, 100, 3), dtype=np.uint8) >>> # Basic usage with default parameters >>> transform = A.Erasing() >>> transformed = transform(image=image) >>> # Custom configuration >>> transform = A.Erasing( ... scale=(0.1, 0.4), ... ratio=(0.5, 2.0), ... fill_value="random_uniform", ... p=1.0 ... ) >>> transformed = transform(image=image)