Public Repository

Last pushed: 3 days ago
Short Description
TensorFlow + Python 2.7 + TensorBoard + JupyterLab + Pandas & more in a docker container for ARMv7
Full Description

Supercharged Machine Learning ToolBox for ARM

A Docker image for ARM devices with Tensorflow 1.7.0 an open source software library for numerical computation using data flow graphs that will let you play and learn distinct Machine Learning techniques over JupyterLab an open-source web application that allows you to create and share documents that contain live code, equations, visualizations and explanatory text. Computational Narratives as the Engine of Collaborative Data Science. All this under Python 2.7 language.
There is very similar image based on Python 3.4 instead of 2.7 elswork/rpi-tensorflow-py3.

Be aware! You should read carefully the usage documentation of every tool!

Latest Changes


  • Upgraded from Tensorflow 1.7.0rc1 to 1.7.0
  • Implementation of the Keras API

1.7.0 rc1

  • Upgraded from Tensorflow 1.6.0 to 1.7.0rc1


  • Upgraded from Tensorflow 1.6.0 rc1 to 1.6.0

1.6.0 rc1

  • Upgraded from Tensorflow 1.5.0 to 1.6.0rc1
  • Replace with new latest binary
  • Replaced base image with ubuntu:16.04 the same as official tensorflow container
  • Change jupyter notebook for jupyterlab


  • Upgraded from Tensorflow 1.5.0 rc1 to 1.5.0
  • Replace with new latest binary

1.5.0 rc1

  • Upgraded from Tensorflow 1.4.0 to 1.5.0rc1
  • Replace with new latest binary



  • Upgraded from Tensorflow 1.2.1 to 1.3.0



Thanks to

Build Instructions

Build for arm32v7 architecture

docker build -t elswork/rpi-tensorflow:latest .

Build for amd64 architecture

docker build -t elswork/rpi-tensorflow:latest \
 --build-arg WHL_URL= \
 --build-arg WHL_FILE=tensorflow-1.7.0rc1-cp27-none-linux_x86_64.whl .

My Real Usage Example

In order everyone could take full advantages of the usage of this docker container, I'll describe my own real usage setup.
For amd64 architecture replace latest by amd64 tag.

docker run -d -p 8888:8888 elswork/rpi-tensorflow:latest

A more complex sample:

docker run -d -p 8888:8888 -p \
 --restart=unless-stopped elswork/rpi-tensorflow:latest

Point your browser to http://localhost:8888

First time you open it, you should provide a Token to log on you cand find it with this command:

docker logs container_name

With the second example you can run TensorBoard executing this command in the container:

tensorboard --logdir=path/to/log-directory --host=

And pointing your browser to http://localhost:6006

Docker Pull Command