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

Andrei Maksimov

Andrei Maksimov


This article will continue the Docker-related article series and show how to create an Ubuntu 20.04 LTS based Docker container for Machine Learning. We’ll install to the environment: Python 3, Jupyter, Keras, Tensorflow, TensorBoard, Pandas, Sklearn, Matplotlib, Seaborn, pyyaml, h5py. Prepare your personal ML environment in 3 minutes, excluding Docker image building time!

Update 2020

I’ve updated the 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 common question when you start learning Machine Learning (ML). In this article, I’ll show you how to create your own Docker container, including the following frameworks for a comfortable start:

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


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

Project Structure

Here’s the final project structure:

$ tree -a python_data_science_container

$ python_data_science_container
├── Dockerfile
├── conf
│   └── .jupyter
│       └──

2 directories, 3 files


All you need to do is to create a project folder and filename 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 always find up-to-date examples 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 its components, it’s time to prepare a configuration for Jupyter. Create a file with the following content:

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

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

Creating startup script

The last thing we need to do is to create a script, which will launch the Jupiter server inside our container during its 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 .

Docker will install all necessary libraries and frameworks inside your container image during the build process 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 Notebooks with the 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 \

It will start the container and expose Jupyter to 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 purposes. I removed authentication on Jupyter in this container, so everybody can connect to port 8888 or 6006 and execute the 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 \

I 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!

How useful was this post?

Click on a star to rate it!

As you found this post useful...

Follow us on social media!

We are sorry that this post was not useful for you!

Let us improve this post!

Tell us how we can improve this post?

Like this article?

Share on facebook
Share on Facebook
Share on twitter
Share on Twitter
Share on linkedin
Share on Linkdin
Share on pinterest
Share on Pinterest

Want to be an author of another post?

We’re looking for skilled technical authors for our blog!

Leave a comment

If you’d like to ask a question about the code or piece of configuration, feel free to use or a similar tool as Facebook comments are breaking code formatting.