Python TensorFlow Lite for amazonlinux

In this repo everything is provided to create the python dependencies nessessary to use tensoflow lite on amazonlinux, which is including support for aws lambda.

Docker

First, if you dont already have it, install Docker. https://www.docker.com/

Dockerfile

Then create this Dockerfile (also available in the repo). The second pip3 install of the dockerfile might change with future versions.

FROM amazonlinux

WORKDIR /tflite

RUN yum groupinstall -y development
RUN yum install -y python3.7
RUN yum install -y python3-devel
RUN pip3 install numpy wheel pybind11

RUN git clone --branch v2.3.0 https://github.com/tensorflow/tensorflow.git
RUN sh ./tensorflow/tensorflow/lite/tools/make/download_dependencies.sh
RUN sh ./tensorflow/tensorflow/lite/tools/pip_package/build_pip_package.sh
RUN pip3 install tensorflow/tensorflow/lite/tools/pip_package/gen/tflite_pip/python3/dist/tflite_runtime-2.3.0-cp37-cp37m-linux_x86_64.whl

CMD tail -f /dev/null

Commands

Run the following commands in the same directory where you created the Dockerfile.

docker build -t tflite_amazonlinux .
docker run -d --name=tflite_amazonlinux tflite_amazonlinux
docker cp tflite_amazonlinux:/usr/local/lib64/python3.7/site-packages .
docker stop tflite_amazonlinux

Site-Packages

In the directory where you ran the commands there should now be a folder called site-packages. In that folder are the correctly compiled tflite python dependencies for amazonlinux. Copy them into your environment in your docker or lambda function and you are good to go.

Warning: These packages will probably not work with other linux distributions or other architectures than amazonlinux. They are specificly created for amazonlinux.

Sources

https://www.tensorflow.org/lite/guide/python https://github.com/tensorflow/tensorflow/tree/master/tensorflow/lite/tools/pip_package

And probably many more :D

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!