tfcoreml - TensorFlow to Core ML Converter

Tool to convert models defined in Tensorflow to the Core ML format.

To get the latest version of tfcoreml from PyPI:

pip install --upgrade tfcoreml
pip install --upgrade coremltools # tfcoreml depends on the coremltools package

For the latest changes please see the release notes.

Usage

Please see the Tensorflow conversion section in the Neural network guide on how to use the converter.

There are several notebook examples as well for reference.

There are scripts in the utils/ directory for visualizing and writing out a text summary of a given frozen TensorFlow graph. This could be useful in determining the input/output names and shapes. Another useful tool for visualizing frozen TensorFlow graphs is Netron.

Dependencies

• tensorflow >= 1.5.0
• coremltools >= 3.1

Installation

Install from Source

To get the latest version of the converter, clone this repository and install from source:

git clone https://github.com/tf-coreml/tf-coreml.git
cd tf-coreml

To install as a package with pip, either run (at the root directory):

pip install -e .

or run:

python setup.py bdist_wheel

This will generate a pip installable wheel inside the dist/ directory.

Install From PyPI

pip install --upgrade tfcoreml

Running Unit Tests

In order to run unit tests, you need pytest.

pip install pytest

To add a new unit test, add it to the tests/ folder. Make sure you name the file with a 'test' as the prefix. To run all unit tests, navigate to the tests/ folder and run

pytest

Directories

• tfcoreml: the tfcoreml package
• examples: examples to use this converter
• tests: unit tests
• utils: general scripts for graph inspection

License

Apache License 2.0