Docker guide

This page only gives element to use and build CARS dockers.

Please go to Docker documentation for more details on Docker technology.

  • First, check Docker install, needed for this page.

$ docker -v


CARS can be used more easily through a docker to avoid the complete installation (with OTB and VLFeat).

  • Install Docker

  • Get public CARS dockerfile images

Main CARS docker image

  • Get official CARS docker image

$ docker pull cnes/cars
$ docker pull cnes/cars:0.5.0 # for a particular cars docker version
  • Run CARS CLI

$ docker run cnes/cars # for CARS command line directly by default "cars -h"

CARS Jupyter notebook docker image

  • Get official CARS Jupyter docker image

$ docker pull cnes/cars-jupyter
$ docker pull cnes/cars-jupyter:0.5.0 # for a particular version
  • Run CARS CLI

$ docker run -p 8888:8888 cnes/cars-jupyter

This runs a jupyter notebook directly to https://localhost:8888/

Follow output indications.

CARS tutorial slideshow

  • Get official CARS tutorial docker image

$ docker pull cnes/cars-tutorial
$ docker pull cnes/cars-tutorial:0.5.0 # for a particular version
  • Run CARS CLI

$ docker run -p 8000:8000 cnes/cars-tutorial

You can go now to https://localhost:8000/


Instead of pulling dockerhub official CARS images, here is the way to build images:

  • Clone CARS repository from GitHub :

$ git clone
$ cd cars
  • Check and build CARS docker image

  $ make docker

The command:

* checks Dockerfiles with hadolint
* builds locally CARS main docker image: cnes/cars
* builds locally CARS jupyter wrapper docker image: cnes/cars-jupyter
* builds locally CARS tutorial wrapper to jupyter docker image: cnes/cars-tutorial

See section docker in Makefile for more details.