How to run Jupiter, Keras, Tensorflow and other ML libs in Docker

Update 2020: I’ve updated container to Ubuntu 20.04 LTS base and speed up Docker build process. Now we’re not building OpenCV from source, but installing it from apt.

Environment setup is a very common question when you’re trying to start learning Machine Learning (ML). In this article I’ll show you how to create your own Docker container including the following frameworks for comfortable start:

That are TOP 10 widely used Python frameworks for Data Science and you’ll find most of them in any HOWTO article on the Internet. In the nex article (How to build Python Data Science Docker container based on Anaconda) I’ll show how to build the same image, but on top of Anaconda distribution.


All you need to have is Docker and text editor installed on your system.


All you need to do, is to create a project folder and file named Dockerfile inside:

mkdir python_data_science_container
cd python_data_science_container
vim Dockerfile

After that put the following content to the Dockerfile:

FROM ubuntu:20.04
MAINTAINER "Andrei Maksimov"

ENV DEBIAN_FRONTEND noninteractive

RUN apt-get update && apt-get install -y \
	libopencv-dev \
        python3-pip \
	python3-opencv && \
    rm -rf /var/lib/apt/lists/*

RUN pip3 install tensorflow && \
    pip3 install numpy pandas sklearn matplotlib seaborn jupyter pyyaml h5py && \
    pip3 install keras --no-deps && \
    pip3 install opencv-python && \
    pip3 install imutils

RUN ["mkdir", "notebooks"]
COPY conf/.jupyter /root/.jupyter

# Jupyter and Tensorboard ports
EXPOSE 8888 6006

# Store notebooks in this mounted directory
VOLUME /notebooks

CMD ["/"]

You may find always up-to-date example of my own Dockerfile on GitHub, which I use to create my personal Data Science container environment (it is also available on Docker Hub for free).

Jupyter configuration

As soon as we declared our container and it’s components, it’s time to prepare a configuration for Jupyter. Create a file with the following content:

c = get_config()  # get the config object
c.IPKernelApp.pylab = 'inline'  # in-line figure when using Matplotlib
c.NotebookApp.ip = '*'
c.NotebookApp.allow_remote_access = True
c.NotebookApp.open_browser = False  # do not open a browser window by default when using notebooks
c.NotebookApp.token = '' # No token. Always use jupyter over ssh tunnel
c.NotebookApp.notebook_dir = '/notebooks'
c.NotebookApp.allow_root = True # Allow to run Jupyter from root user inside Docker container

As you can guess from Dockerfile, we’ll put it to /root/.jupyter/ folder during container build process.

Creating startup script

The last thing we need to do is to create a script, which will launch Jupiter server inside our container during it’s starting process. Create a with the following content:

#!/usr/bin/env bash

jupyter notebook "$@"

And make this file executable:

chmod +x

This file will be launched inside your container by default each time you’ll start the new one.

Creating container image

The last stage – container creation. Just run the following command to build your Docker container from the project directory:

docker build -f Dockerfile -t python_data_science_container .

During build process Docker will install all necessary libraries and frameworks inside your container image and make it available for use.

Running container

Now you have a working container and it’s time to start it. Create a folder inside your project’s folder where we’ll store all our Jupyter Noteboos with source code of our projects:

mkdir notebooks

And start the container with the following command:

docker run -it -p 8888:8888 -p 6006:6006 -d -v $(pwd)/notebooks:/notebooks python_data_science_container

It will start the container and expose Jupyter on port 8888 and Tensorflow Dashboard on port 6006 on your local computer or your server depending on where you’re executed this command.

Please be aware that this container was created only for local development purpose and I removed authentication on Jupyter in this container, so everybody can connect to port 8888 or 6006 and execute Python code of cause.

If you’re just looking for a working solution

If you don’t want to create and maintain your own container and aforesaid components will be sufficient for you, please feel free to use my personal container, that I usually update:

docker run -it -p 8888:8888 -p 6006:6006 -d -v $(pwd)/notebooks:/notebooks amaksimov/python_data_science

Hope, this article will be helpful for you. If you like the article, please repost it using any social media you’d like. See you soon!


Author avatar
Andrei Maksimov

I’m a passionate Cloud Infrastructure Architect with more than 15 years of experience in IT.

Let’s discuss your AWS questions if you still have them.

This post represents my personal experience and opinion about the topic.