Albumentations 1.4.24 Release Notes

• Support Our Work
• Core
• Transforms
• Bugfixes

Support Our Work

1. Help Us Grow - If you find value in Albumentations, consider becoming a sponsor. Every contribution, no matter the size, helps us maintain and improve the library for everyone.
2. Show Your Support - If you enjoy using Albumentations, consider giving us a ⭐ on GitHub. It helps others discover the library and motivates our team.
3. Join Our Community - Have suggestions or ran into issues? We welcome your input! Share your experience in our GitHub issues or connect with us on Discord.

Core

• Added new keypoints format xyz for ImageOnly and Dual transforms (z coordinate stays unchanged)

Transforms

New transform AtLeastOneBBoxRandomCrop

Crop an area from image while ensuring at least one bounding box is present in the crop.

by @guillaume-rochette-oxb

Improvements

• SmallestMaxSize: Added option for separate max_size for height/width
• LongestMaxSize: Added option for separate max_size for height/width
• Added keypoints support to: CenterCrop3D, CoarseDropout3D, CubicSymmetry, Pad3D, PadIfNeeded3D, RandomCrop3D (by @ternaus)

Bugfixes

• Do not import eval-type-backport for python 3.10 and older. by @PerchunPak
• Bugfix in ToTensorV2 by @matejpekar

Albumentations 1.4.23 Release Notes

• Support Our Work
• Core
• Transforms
• Bugfixes

Support Our Work

1. Help Us Grow - If you find value in Albumentations, consider becoming a sponsor. Every contribution, no matter the size, helps us maintain and improve the library for everyone.
2. Show Your Support - If you enjoy using Albumentations, consider giving us a ⭐ on GitHub. It helps others discover the library and motivates our team.
3. Join Our Community - Have suggestions or ran into issues? We welcome your input! Share your experience in our GitHub issues or connect with us on Discord.

Core

Target images as numpy array

Now supports numpy arrays with shape (num_images, height, width, num_channels) or (num_images, height, width) as images in Compose

• Ideal for video processing applications
• Same transform applies to all images in the array

New 3D Data Support

• volume: (depth, height, width) or (depth, height, width, num_channels)
• mask3d: (depth, height, width) or (depth, height, width, num_channels)
• volumes: (num_volumes, depth, height, width) for batch processing
• masks3d: (num_volumes, depth, height, width) for batch processing

volume = np.random.rand(96, 256, 256) # Your 3D medical volume
mask = np.zeros((96, 256, 256)) # Your 3D segmentation mask
transformed = transform(volume=volume, mask3d=mask)
transformed_volume = transformed['volume']
transformed_mask = transformed['mask3d']

Transforms

Added 3D transforms by @ternaus

Padding & Cropping

• Pad3D: Pad 3D volumes with flexible padding options
• PadIfNeeded3D: Conditional padding to meet minimum dimensions or divisibility requirements
• CenterCrop3D: Center cropping for 3D volumes
• RandomCrop3D: Random cropping of 3D volumes

transform = A.Compose([
    # Crop volume to a fixed size for memory efficiency
    A.RandomCrop3D(size=(64, 128, 128), p=1.0),    
    # Randomly remove cubic regions to simulate occlusions
    A.CoarseDropout3D(
        num_holes_range=(2, 6),
        hole_depth_range=(0.1, 0.3),
        hole_height_range=(0.1, 0.3),
        hole_width_range=(0.1, 0.3),
        p=0.5
    ),    
])

volume = np.random.rand(96, 256, 256) # Your 3D medical volume
mask = np.zeros((96, 256, 256)) # Your 3D segmentation mask
transformed = transform(volume=volume, mask3d=mask)
transformed_volume = transformed['volume']
transformed_mask = transformed['mask3d']

Augmentation

• CoarseDropout3D: Random cuboid dropout regions for occlusion simulation
• CubicSymmetry: 48 possible cube symmetry transformations (24 rotations + 24 rotoreflections)

Fixes

• Added flexible brightness in RandomSunFlare by @momincks
• Bugfix in CenterCrop, RandomCrop by @iRyoka
• Fix in Normalize docstring by @mennohofste

Albumentations 1.4.22 Release Notes

• Support Our Work
• Transforms
• Core
• Bugfixes

Support Our Work

1. Help Us Grow - If you find value in Albumentations, consider becoming a sponsor. Every contribution, no matter the size, helps us maintain and improve the library for everyone.
2. Show Your Support - If you enjoy using Albumentations, consider giving us a ⭐ on GitHub. It helps others discover the library and motivates our team.
3. Join Our Community - Have suggestions or ran into issues? We welcome your input! Share your experience in our GitHub issues or connect with us on Discord.

Transforms

Elastic Transform

1. Added argument noise_distribution that allows sampling displacement fields from gaussian and from uniform distributions.
2. Deprecated parameters border_mode, value, mask_value - you can specify them, but will not have any effect.

New transform ShotNoise

Apply shot noise to the image by modeling photon counting as a Poisson process.

    Shot noise (also known as Poisson noise) occurs in imaging due to the quantum nature of light.
    When photons hit an imaging sensor, they arrive at random times following Poisson statistics.
    This transform simulates this physical process in linear light space by:
    1. Converting to linear space (removing gamma)
    2. Treating each pixel value as an expected photon count
    3. Sampling actual photon counts from a Poisson distribution
    4. Converting back to display space (reapplying gamma)

    The noise characteristics follow real camera behavior:
    - Noise variance equals signal mean in linear space (Poisson statistics)
    - Brighter regions have more absolute noise but less relative noise
    - Darker regions have less absolute noise but more relative noise
    - Noise is generated independently for each pixel and color channel

RandomGridShuffle

Addes support for bounding boxes

CorseDropout

Added an option to inpaint holes using inpaint_ns and inpaint_telea from OpenCV

GridDropout

Added an option to inpaint holes using inpaint_ns and inpaint_telea from OpenCV

MaskDropout

Added an option to inpaint holes using inpaint_ns and inpaint_telea from OpenCV

XYMasking

Added an option to inpaint holes using inpaint_ns and inpaint_telea from OpenCV

New transform TimeReverse

Added NewTransform TimeReverse

Reverse the time axis of a spectrogram image, also known as time inversion.

    Time inversion of a spectrogram is analogous to the random flip of an image,
    an augmentation technique widely used in the visual domain. This can be relevant
    in the context of audio classification tasks when working with spectrograms.
    The technique was successfully applied in the AudioCLIP paper, which extended
    CLIP to handle image, text, and audio inputs.

    This transform is implemented as a subclass of HorizontalFlip since reversing
    time in a spectrogram is equivalent to flipping the image horizontally.

New transform TimeMasking

Added NewTransform TimeMasking

Apply masking to a spectrogram in the time domain.

    This transform masks random segments along the time axis of a spectrogram,
    implementing the time masking technique proposed in the SpecAugment paper.
    Time masking helps in training models to be robust against temporal variations
    and missing information in audio signals.

    This is a specialized version of XYMasking configured for time masking only.
    For more advanced use cases (e.g., multiple masks, frequency masking, or custom
    fill values), consider using XYMasking directly.

New transform FrequencyMasking

Apply masking to a spectrogram in the frequency domain.

    This transform masks random segments along the frequency axis of a spectrogram,
    implementing the frequency masking technique proposed in the SpecAugment paper.
    Frequency masking helps in training models to be robust against frequency variations
    and missing spectral information in audio signals.

    This is a specialized version of XYMasking configured for frequency masking only.
    For more advanced use cases (e.g., multiple masks, time masking, or custom
    fill values), consider using XYMasking directly.

Added NewTransform FrequencyMasking

It is a specialized version of XYMasking that has the similar API as FrequencyMasking from torchaudio

New Transform Pad

Pad the sides of an image by specified number of pixels.

    Args:
        padding (int, tuple[int, int] or tuple[int, int, int, int]): Padding values. Can be:
            * int - pad all sides by this value
            * tuple[int, int] - (pad_x, pad_y) to pad left/right by pad_x and top/bottom by pad_y
            * tuple[int, int, int, int] - (left, top, right, bottom) specific padding per side

This is the generalization of the torchvision transform with the same name

New Transform Erasing

This is the generalization of the similar torchvision transform

Randomly erases rectangular regions in an image, following the Random Erasing Data Augmentation technique.

    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.

New Transform AdditiveNoise

Apply random noise to image channels using various noise distributions.

    This transform generates noise using different probability distributions and applies it
    to image channels. The noise can be generated in three spatial modes and supports
    multiple noise distributions, each with configurable parameters.

    Args:
        noise_type: Type of noise distribution to use. Options:
            - "uniform": Uniform distribution, good for simple random perturbations
            - "gaussian": Normal distribution, models natural random processes
            - "laplace": Similar to Gaussian but with heavier tails, good for outliers
            - "beta": Flexible bounded distribution, can be symmetric or skewed

        spatial_mode: How to generate and apply the noise. Options:
            - "constant": One noise value per channel, fastest
            - "per_pixel": Independent noise value for each pixel and channel, slowest
            - "shared": One noise map shared across all channels, medium speed

Sharpen

Added 'gaussian' method for image sharpening.

New transform SaltAndPepper

Apply salt and pepper noise to the input image.

    Salt and pepper noise is a form of impulse noise that randomly sets pixels to either maximum value (salt)
    or minimum value (pepper). The amount and proportion of salt vs pepper noise can be controlled.

New transform PlasmaBrightNessContrast

Apply plasma fractal pattern to modify image brightness and contrast.

    This transform uses the Diamond-Square algorithm to generate organic-looking fractal patterns
    that are then used to create spatially-varying brightness and contrast adjustments.
    The result is a natural-looking, non-uniform modification of the image.

New Transform PlasmaShadow

<img width="118...

Albumentations 1.4.21 Release Notes

• Support Our Work
• Transforms
• Core
• Benchmark
• Speedups

Support Our Work

1. Love the library? You can contribute to its development by becoming a sponsor for the library. Your support is invaluable, and every contribution makes a difference.
2. Haven't starred our repo yet? Show your support with a ⭐! It's just only one mouse click away.
3. Got ideas or facing issues? We'd love to hear from you. Share your thoughts in our issues or join the conversation on our Discord server

Transforms

Auto padding in crops

Added option to pad the image if crop size is larger than the crop size

Old way

[
A.PadIfNeeded(min_height=1024, min_width=1024, p=1),
A.RandomCrop(height=1204, width=1024, p=1)
]

New way:

A.RandomCrop(height=1204, width=1024, p=1, pad_if_needed=True)

Works for:

• RandomCrop
• CenterCrop
• Crop

You may also use it to pad image to a desired size.

Core

Random state

Now random state for the pipeline does not depend on the global random state

Before

random.seed(seed)
np.random.seed(seed)

transform = A.Compose(...) 

Now

transform = A.Compose(seed=seed, ...)

or

transform = A.Compose(...)
transform.set_random_seed(seed)

Saving used parameters

Now you can get exact parameters that were used in the pipeline on a given sample with

transform = A.Compose(save_applied_params=True, ...)

result = transform(image=image, bboxes=bboxes, mask=mask, keypoints=keypoints)

print(result["applied_transforms"])

Benchmark

Moved benchmark to a separate repo

https://github.com/albumentations-team/benchmark/

Current result for uint8 images:

Transform	albumentations 1.4.20	augly 1.0.0	imgaug 0.4.0	kornia 0.7.3	torchvision 0.20.0
HorizontalFlip	8325 ± 955	4807 ± 818	6042 ± 788	390 ± 106	914 ± 67
VerticalFlip	20493 ± 1134	9153 ± 1291	10931 ± 1844	1212 ± 402	3198 ± 200
Rotate	1272 ± 12	1119 ± 41	1136 ± 218	143 ± 11	181 ± 11
Affine	967 ± 3	-	774 ± 97	147 ± 9	130 ± 12
Equalize	961 ± 4	-	581 ± 54	152 ± 19	479 ± 12
RandomCrop80	118946 ± 741	25272 ± 1822	11503 ± 441	1510 ± 230	32109 ± 1241
ShiftRGB	1873 ± 252	-	1582 ± 65	-	-
Resize	2365 ± 153	611 ± 78	1806 ± 63	232 ± 24	195 ± 4
RandomGamma	8608 ± 220	-	2318 ± 269	108 ± 13	-
Grayscale	3050 ± 597	2720 ± 932	1681 ± 156	289 ± 75	1838 ± 130
RandomPerspective	410 ± 20	-	554 ± 22	86 ± 11	96 ± 5
GaussianBlur	1734 ± 204	242 ± 4	1090 ± 65	176 ± 18	79 ± 3
MedianBlur	862 ± 30	-	813 ± 30	5 ± 0	-
MotionBlur	2975 ± 52	-	612 ± 18	73 ± 2	-
Posterize	5214 ± 101	-	2097 ± 68	430 ± 49	3196 ± 185
JpegCompression	845 ± 61	778 ± 5	459 ± 35	71 ± 3	625 ± 17
GaussianNoise	147 ± 10	67 ± 2	206 ± 11	75 ± 1	-
Elastic	171 ± 15	-	235 ± 20	1 ± 0	2 ± 0
Clahe	423 ± 10	-	335 ± 43	94 ± 9	-
CoarseDropout	11288 ± 609	-	671 ± 38	536 ± 87	-
Blur	4816 ± 59	246 ± 3	3807 ± 325	-	-
ColorJitter	536 ± 41	255 ± 13	-	55 ± 18	46 ± 2
Brightness	4443 ± 84	1163 ± 86	-	472 ± 101	429 ± 20
Contrast	4398 ± 143	736 ± 79	-	425 ± 52	335 ± 35
RandomResizedCrop	2952 ± 24	-	-	287 ± 58	511 ± 10
Normalize	1016 ± 84	-	-	626 ± 40	519 ± 12
PlankianJitter	1844 ± 208	-	-	813 ± 211	-

Speedups

• Speedup in PlankianJitter in uint8 mode
• Replaced cv2.addWeighted with wsum from simsimd package

Albumentations 1.4.20 Release Notes

Hotfix version.

• Fix in check_version
• Fix in PieceWiseAffine
• Fix in RandomSizedCrop and RandomResizedCrop
• Fix in RandomOrder

Albumentations 1.4.19 Release Notes

• Support Our Work
• Transforms
• Core
• Bug Fixes

Support Our Work

1. Love the library? You can contribute to its development by becoming a sponsor for the library. Your support is invaluable, and every contribution makes a difference.
2. Haven't starred our repo yet? Show your support with a ⭐! It's just only one mouse click away.
3. Got ideas or facing issues? We'd love to hear from you. Share your thoughts in our issues or join the conversation on our Discord server

Transforms

Added mask_interpolation to all transforms that use mask interpolation, including:

• RandomSizedCrop
• RandomResizedCrop
• RandomSizedBBoxSafeCrop
• CropAndPad
• Resize
• RandomScale
• LongestMaxSize
• SmallestMaxSize
• Rotate
• SafeRotate
• OpticalDistortion
• GridDistortion
• ElasticTransform
• Perspective
• PiecewiseAffine

by @ternaus

Core

• Minimal supported python version is 3.9
• Removed dependency on scikit-image
• Updated Random number generator from np.random.state to np.random.generator. Second is 50% faster => speedups in all transforms that heavily use random generator
• Where possible moved from cv2.LUT to stringzilla lut
• Added parameter mask_interpolation to Compose that overrides mask interpolation value in all transforms in that Compose, now can use more accurate cv2.INTER_NEAREST_EXACT for semantic segmentation and can work with depth and heatmap estimation using cubic, area, linear, etc

BugFixes

• Bugfix in ISONoise
• Bugfix: Ensure that transforms masks are contiguous arrays, by @Callidior
• Bugfix in Solarize
• Bugfix in bounding box filtering
• Bugfix in OpticalDistortion
• Bugfix in balanced scale in Affine

Albumentations 1.4.18 Release Notes

• Support Our Work
• Transforms
• Core
• Deprecations
• Bugfixes

Support Our Work

1. Love the library? You can contribute to its development by becoming a sponsor for the library. Your support is invaluable, and every contribution makes a difference.
2. Haven't starred our repo yet? Show your support with a ⭐! It's just only one mouse click away.
3. Got ideas or facing issues? We'd love to hear from you. Share your thoughts in our issues or join the conversation on our Discord server

Transforms

GridDistortion

Added support for keypoints

GridDropout

Added support for keypoints and bounding boxes

GridElasticDeform

Added support for keypoints and bounding boxes

MaskDropout

Added support for keypoints and bounding boxes

Morphological

Added support for bounding boxes and keypoints

OpticalDistortion

Added support for keypoints

PixelDropout

Added support for keypoints and bonding boxes

XYMasking

Added support for bounding boxes and keypoints

Core

Added support for masks as numpy arrays of the shape (num_masks, height, width)

Now you can apply transforms to masks as:

masks = <numpy array with shape (num_masks, height, width)>

transform(image=image, masks=masks)

Deprecations

Removed MixUp as it was doing almost exactly the same as TemplateTransform

Bugfixes

• Bugfix in RandomFog
• Bugfix in PlankianJitter
• Several people reported issue with masks as list of numpy arrays, I guess it was fixed as a part of some other work as I cannot reproduce it. Just in case added tests for that case.

Albumentations 1.4.17 Release Notes

• Support Our Work
• Transforms
• Core

Support Our Work

1. Love the library? You can contribute to its development by becoming a sponsor for the library. Your support is invaluable, and every contribution makes a difference.
2. Haven't starred our repo yet? Show your support with a ⭐! It's just only one mouse click away.
3. Got ideas or facing issues? We'd love to hear from you. Share your thoughts in our issues or join the conversation on our Discord server

Transforms

CoarseDropout

1. Added Bounding Box support
2. remove_invisible=False keeps keypoints

by @ternaus

ElasticTransform

Added support for keypoints

by @ternaus

Core

Added RandomOrder Compose

Select N transforms to apply. Selected transforms will be called in random order with force_apply=True. 
Transforms probabilities will be normalized to one 1, so in this case transforms probabilities works as weights. 
This transform is like SomeOf, but transforms are called with random order. 
It will not replay random order in ReplayCompose.

Albumentations 1.4.16 Release Notes

• Support Our Work
• UI Tool
• Transforms
• Improvements and Bug Fixes

Support Our Work

1. Love the library? You can contribute to its development by becoming a sponsor for the library. Your support is invaluable, and every contribution makes a difference.
2. Haven't starred our repo yet? Show your support with a ⭐! It's just only one mouse click away.
3. Got ideas or facing issues? We'd love to hear from you. Share your thoughts in our issues or join the conversation on our Discord server

UI Tool

For visual debug wrote a tool that allows visually inspect effects of augmentations on the image.

You can find it at https://explore.albumentations.ai/

• Works for all ImageOnly transforms
• Authorized users can upload their own images

it is work in progress. It is not stable and polished yet, but if you have feedback or proposals - just write in the Discord Server mentioned above.

Transforms

• Updated and extended docstrings in all ImageOnly transforms.
• All ImageOnly transforms support both uint8 and float32 inputs

RandomSnow

Added texture method to RandomSnow

RandomSunflare

Added physics_based method to RandomSunFlare

Bugfixes and improvements

• Bugfix in albucore dependency. Now every Albumnetations version is tailored to a specific albucore version. Added pre-commit hook to automatically check it on every commit.
• BugFix in TextImage transform, after rewriting bbox processing in a vectorized form, transform was failing.
• As a part of the work to remove scikit-image dependency @momincks rewrote bbox_affine in a plain numpy
• Bugfix. It was unexpected, but people use bounding bboxes that are less than 1 pixel. Removed constrant on a minimum bounding box being 1x1
• Bugfix in bounding box filtering. Now if all bounding boxes were filtered return not empty array, but empty array of shape (0, 4)

Albumentations 1.4.15 Release Notes

• Support Our Work
• UI Tool
• Core
• Transforms
• Improvements and Bug Fixes

Support Our Work

1. Love the library? You can contribute to its development by becoming a sponsor for the library. Your support is invaluable, and every contribution makes a difference.
2. Haven't starred our repo yet? Show your support with a ⭐! It's just only one mouse click away.
3. Got ideas or facing issues? We'd love to hear from you. Share your thoughts in our issues or join the conversation on our Discord server

UI Tool

For visual debug wrote a tool that allows visually inspect effects of augmentations on the image.

You can find it at https://explore.albumentations.ai/

RIght now supports only ImageOnly transforms, and not all but a subset of them.

it is work in progress. It is not stable and polished yet, but if you have feedback or proposals - just write in the Discord Server mentioned above.

Core

Bounding box and keypoint processing was vectorized

• You can pass numpy array to compose and not only list of lists.
• Presumably transforms will work faster, but did not benchmark.

Transforms

Affine

• Reflection padding correctly works In Affine and ShiftScaleRotate

CLAHE

• Added support for float32 images

Equalize

• Added support for float32 images

FancyPCA

• Added support for float32 images
• Added support for any number of channels

PixelDistributionAdaptation

• Added support for float32
• Added support for anyu number of channels

Flip

Still works, but deprecated. It was a very strange transform, I cannot find use case, where you needed to use it.

It was equivalent to:

OneOf([Transpose, VerticalFlip, HorizontalFlip]) 

Most likely if you needed transform that does not create artifacts, you should look at:

• Natural images => HorizontalFlip (Symmetry group has 2 elements, meaning will effectively increase your dataset 2x)
• Images that look natural when you vertically flip them => VerticalFlip (Symmetry group has 2 elements, meaning will effectively increase your dataset 2x)
• Images that need to preserve parity, for example texts, but we may expect rotated documents => RandomRotate90 (Symmetry group has 2 elements, meaning will effectively increase your dataset 4x)
• Images that you can flip and rotate as you wish => D4 (Symmetry group has 8 elements, meaning will effectively increase your dataset 8x)

ToGray

Now you can define the number of output channels in the resulting gray image. All channels will be the same.

Extended ways one can get grayscale image. Most of them can work with any number of channels as input

• weighted_average: Uses a weighted sum of RGB channels (0.299R + 0.587G + 0.114B)
  Works only with 3-channel images. Provides realistic results based on human perception.
• from_lab: Extracts the L channel from the LAB color space.
  Works only with 3-channel images. Gives perceptually uniform results.
• desaturation: Averages the maximum and minimum values across channels.
  Works with any number of channels. Fast but may not preserve perceived brightness well.
• average: Simple average of all channels.
  Works with any number of channels. Fast but may not give realistic results.
• max: Takes the maximum value across all channels.
  Works with any number of channels. Tends to produce brighter results.
• pca: Applies Principal Component Analysis to reduce channels.
  Works with any number of channels. Can preserve more information but is computationally intensive.

SafeRotate

Now uses Affine under the hood.

Improvements and Bug Fixes

• Bugfix in GridElasticDeform by @4pygmalion
• Speedups in to_float and from_float
• Bugfix in PadIfNeeded. Did not work when empty bounding boxes were passed.