Multidimensional (2D and 3D) Image Augmentation for TensorFlow

Note: this package is archived and no longer actively maintained on github. It remains as a public snapshot of the internal project as of January 2021.

This package provides TensorFlow Ops for multidimensional volumetric image augmentation.

Install prerequities

This project usings the bazel build system

Build and test

To fetch the code, build it, and run tests:

git clone https://github.com/deepmind/multidim-image-augmentation.git
cd multidim-image-augmentation/
bazel test --python_version=py3 --config=nativeopt //...

Note bazel 0.24.0 made a lot of backward incompatible changes to default flag values, that have not yet been resolved in this project and its dependencies. In the meantime, you can disable with a few simple extra flags:

# Bazel >= 0.24.0
bazel test \
    --incompatible_disable_genrule_cc_toolchain_dependency=false \
    --incompatible_disable_legacy_cc_provider=false \
    --incompatible_disable_third_party_license_checking=false \
    --incompatible_no_transitive_loads=false \
    --incompatible_bzl_disallow_load_after_statement=false \
    --incompatible_disallow_load_labels_to_cross_package_boundaries=false \
    --config=nativeopt //...

To learn more about image augmentation, see the primer

For simple API usage examples, see the python test code.