Python Course
Welcome to the Python part of the Robot Vision Systems course.
Preparations
In the course we will use Python 3. If possible, get the latest version 3.5, otherwise 3.4 will be fine. Earlier versions are not recommended.
You will also need to install the python packages scipy, numpy and matplotlib. Since the course is based on OpenCV, you need to install the Python bindings for that too. These are all more or less easy to install, depending on your platform. For Linux systems installing via your distributions package manager (i.e. yum/dnf, apt-get, etc.) is usually quite easy. For Mac and Windows it is apparently often much more difficult. Therefore I recommend you to use a stand-alone Python distribution instead. My recommendation is the Anaconda Python distribution.
Anaconda
Using Anaconda is IMHO the easiest way to use and install the Python scientific computing stack. Instead of the full anaconda install, you can also choose to use the miniconda version, and only install the packages that you actually need. Note that the Anaconda installer says "Python 3.4", which is fine, since you can still use the conda package manager to install a newer python version.
Do not allow anaconda to become the default python interpreter! At least on Linux and Mac systems this is likely to give you all sorts of problems.
There is a short tutorial you can follow, to test your install of anaconda/miniconda.
After installing anaconda or miniconda, you should use the conda
command
to create a suitable environment:
$ conda create --name pycourse python=3.5 numpy=1.10 scipy h5py pandas ipython ipython-notebook
matplotlib
This creates a virtual python environment named "pycourse" with the packages that I will use for the course.
The environment can be activated by running
source activate pycourse
on Linux and Mac and
activate pycourse
on Windows.
To test the environment, just run python --version
and make sure
the output looks similar to
Python 3.5.0 :: Continuum Analytics, Inc.
OpenCV
If you chose to use anaconda, then you can install a version of opencv 3.0 compatible with Python 3.5 and numpy 1.10 using
$ conda install -c hovren opencv
For recent versions of Fedora (22) you will then have to remove the fontconfig conda package, to avoid errors when importing the cv2 module:
$ conda remove fontconfig
Assignments
Lectures
The lecture slides are available both as rendered HTML, and as the original
IPython interactive notebook.
To run the notebooks simply download the files and run ipython notebook
in that directory.
- Lecture 1 - Python introduction [Slides] [Notebook]
- Lecture 2 - Scientific Python [Slides] [Notebook]
- Lecture 3 - Mixed topics [Slides] [Notebook]
Informationsansvarig: Hannes Ovrén
Senast uppdaterad: 2015-11-18