Release 1.3

Release 1.3 is a maintenance release. No new features were added.

This release requires TensorFlow 1.15.

If you're on Linux or Darwin (Mac OS), install the pre-compiled binary by running:

pip install tensorflow-compression==1.3

Release 1.2

Release 1.2 is the first major feature upgrade for tensorflow-compression. Because it contains custom TensorFlow ops, you need to compile this release to use it. If you're on Linux or Darwin (Mac OS), install the pre-compiled binary by running:

pip install tensorflow-compression==1.2

This release requires TensorFlow 1.14.

New features in 1.2:

• Proper release schedule and binaries. We've tagged existing code versions 1.0 and 1.1, which work without installing binaries with TensorFlow <=1.12 and 1.13, respectively. These versions rely on a binary range coder implementation in tf.contrib, which will not exist any more in TensorFlow 1.14 (the TensorFlow team will remove tf.contrib entirely in release 2.0). For this reason, tensorflow-compression 1.2 ships with its own range coder implementations. Pre-compiled pip packages are provided for Linux and Darwin (Mac).
• Support for conditional entropy models. This version includes three new classes, GaussianConditional,LaplacianConditional, and LogisticConditional, which implement conditional entropy models (entropy models whose probabilities are computed using another neural network). These are necessary for building the hyperprior models published in our ICLR 2018 paper. This also includes a more flexible range coding implementation, which can encode unbounded integer values. An example training script for the ICLR 2018 model is provided in the examples directory.
• Support for trained models. The library now includes support for a new file format, TFCI, and packaged metagraphs of models we've published. examples/tfci.py implements an easy-to-use command line interface for converting PNG images to TFCI and back.
• More flexible SignalConv layers. Among some other minor improvements, the SignalConv* Keras layers now implement rational up-/downsampling factors.

Beta 2 for release 1.2

Release 1.2 is the first major feature upgrade for tensorflow-compression. Because it contains custom TensorFlow ops, you need to compile this release to use it. If you're on Linux or Darwin (Mac OS), install the pre-compiled binary by running:

pip install tensorflow-compression==1.2b2

This beta release requires TensorFlow 1.14.

New features in 1.2:

• Proper release schedule and binaries. We've tagged existing code versions 1.0 and 1.1, which work without installing binaries with TensorFlow <=1.12 and 1.13, respectively. These versions rely on a binary range coder implementation in tf.contrib, which will not exist any more in TensorFlow 1.14 (the TensorFlow team will remove tf.contrib entirely in release 2.0). For this reason, tensorflow-compression 1.2 ships with its own range coder implementations. Pre-compiled pip packages are provided for Linux and Darwin (Mac).
• Support for conditional entropy models. This version includes three new classes, GaussianConditional,LaplacianConditional, and LogisticConditional, which implement conditional entropy models (entropy models whose probabilities are computed using another neural network). These are necessary for building the hyperprior models published in our ICLR 2018 paper. This also includes a more flexible range coding implementation, which can encode unbounded integer values.
• Support for trained models. The library now includes support for a new file format, TFCI, and packaged metagraphs of models we've published. examples/tfci.py implements an easy-to-use command line interface for converting PNG images to TFCI and back.
• More flexible SignalConv layers. Among some other minor improvements, the SignalConv* Keras layers now implement rational up-/downsampling factors.

Upcoming features not yet implemented in this beta:

• Example file for training your own model with a hierarchical prior (as in the ICLR 2018 paper).
• More (and better) documentation.

Beta 1 for release 1.2

Release 1.2 is the first major feature upgrade for tensorflow-compression. Because it contains custom TensorFlow ops, you need to compile this release to use it. If you're on Linux, install the pre-compiled binary by running:

pip install tensorflow-compression==1.2b1

This beta release requires TensorFlow 1.13.

New features in 1.2:

• Proper release schedule and binaries. We've tagged existing code versions 1.0 and 1.1, which work without installing binaries with TensorFlow <=1.12 and 1.13, respectively. These versions rely on a binary range coder implementation in tf.contrib, which will not exist any more in TensorFlow 1.14 (the TensorFlow team will remove tf.contrib entirely in release 2.0). For this reason, tensorflow-compression 1.2 ships with its own range coder implementations. Pre-compiled pip packages will be provided for Linux initially. We are working with the TensorFlow team to also provide Darwin (Mac) binaries.
• Support for conditional entropy models. This version includes three new classes, GaussianConditional,LaplacianConditional, and LogisticConditional, which implement conditional entropy models (entropy models whose probabilities are computed using another neural network). These are necessary for building the hyperprior models published in our ICLR 2018 paper. This also includes a more flexible range coding implementation, which can encode unbounded integer values.
• Support for trained models. The library now includes support for a new file format, TFCI, and packaged metagraphs of models we've published. examples/tfci.py implements an easy-to-use command line interface for converting PNG images to TFCI and back.
• More flexible SignalConv layers. Among some other minor improvements, the SignalConv* Keras layers now implement rational up-/downsampling factors.

Upcoming features not yet implemented in this beta:

• More pre-trained models (equivalent to publications End-to-end optimized image compression as well as Joint Autoregressive and Hierarchical Priors for Learned Image Compression).
• Example file for training your own model with a hierarchical prior (as in the ICLR 2018 paper).
• Native Darwin (Mac) binaries. We can't promise we'll get support ready for 1.2, but we will try.
• More (and better) documentation.

Release 1.1

This release does not need to be compiled, and was tested with TensorFlow 1.13.

You can install this release simply by downloading the ZIP file, or checking out the tagged commit v1.1 with git. Note: this release is incompatible with TensorFlow >=1.14.

Release 1.0

This release does not need to be compiled, and was tested with TensorFlow 1.12.

You can install this release simply by downloading the ZIP file, or checking out the tagged commit v1.0 with git. Note: this release is incompatible with TensorFlow >=1.13.