Skip to content

Frequently Asked Questions

Installation

Examples

Usage

Installation

I am receiving an error message Failed building wheel for imagecodecs when I am trying to install Albumentations. How can I fix the problem?

Try to update pip by running the following command:

Bash
python3 -m pip install --upgrade pip

I successfully installed the library, but when I am trying to import it I receive an error ImportError: libXrender.so.1: cannot open shared object file: No such file or directory.

Probably your system doesn't have libXrender. To install the libXrender package on Ubuntu or Debian run:

Bash
 sudo apt-get update
 sudo apt-get install libxrender1

To install the package on other operating systems, consult the documentation for the OS' package manager.

Examples

Why do you call cv2.cvtColor(image, cv2.COLOR_BGR2RGB) in your examples?

For historical reasons, OpenCV reads an image in BGR format (so color channels of the image have the following order: Blue, Green, Red). Albumentations uses the most common and popular RGB image format. So when using OpenCV, we need to convert the image format to RGB explicitly.

Usage

How can I find which augmentations were applied to the input data and which parameters they used?

To save and inspect parameters of augmentations, you can replace Compose with ReplayCompose. ReplayCompose behaves just like regular Compose, but it also saves information about which augmentations were applied and which parameters were uses. Take a look at the example that shows how you can use ReplayCompose.

My computer vision pipeline works with a sequence of images. I want to apply the same augmentations with the same parameters to each image in the sequence. Can Albumentations do it?

Yes. You can define additional images, masks, bounding boxes, or keypoints through the additional_targets argument to Compose. You can then pass those additional targets to the augmentation pipeline, and Albumentations will augment them in the same way. See this example for more info.

I want to augment 16-bit TIFF images. Can Albumentations work with them?

Yes. Albumentations can also work with non-8-bit images. See this example for more info.

Augmentations have a parameter named p that sets the probability of applying that augmentation, but they also have the always_apply parameter that can either be True or False. What is the difference between p and always_apply? Is always_apply=True equals to p=1.0?

When always_apply is set to True, Albumentations will always apply that transform, even if p is set to a value less than 1.0. However, always_apply=True doesn't equal to p=1.0 because with always_apply=True, Albumentations will apply a transform even in a case when top-level containers are not applied.

Let's look at an example when a container Compose contains one augmentation Resize:

Python
transform = A.Compose([
    A.Resize(height=256, width=256, p=1.0),
], p=0.9)

If you set p=1.0 for Resize and p=0.9 for Compose, then Resize has a 90% chance to be applied, because there is a 90% chance for Compose to be applied and if Compose is applied, there is a 100% chance for Resize to be applied.

But if you set always_apply=True for Resize, Albumentations will apply it with 100% probability even if Albumentations decides not to apply the parent container (Compose in the example):

Python
transform = A.Compose([
    A.Resize(height=256, width=256, always_apply=True),
], p=0.9)

When I use augmentations with the border_mode parameter (such as Rotate) and set border_mode to cv2.BORDER_REFLECT or cv2.BORDER_REFLECT_101 Albumentations mirrors regions of images and masks but doesn't mirror bounding boxes and keypoints. Is it a bug?

Unfortunately, adding extra bounding boxes or keypoints to reflected regions of the image is not supported. You can change border_mode mode to cv2.BORDER_CONSTANT if this causes a significant impact on the training of your model.

I created annotations for bounding boxes using labeling service or labeling software. How can I use those annotations in Albumentations?

You need to convert those annotations to one of the formats, supported by Albumentations. For the list of formats, please refer to this article. Consult the documentation of the labeling service to see how you can export annotations in those formats.