ilastik Installation


From binary

We recommend to install the Missing page "ilastik 0.5 binaries" in category "Download" environment of ilastik. This enables you to use ilastik without taking care of all dependency packages. The binary version bundles all required dependencies and the source code of ilastik is also included. If, however, you want to add custom functionality or contribute to the ilastik code base you probably want to install ilastik from source.

From source

In order to install ilastik from source you will have to install the version control system git, which is part of any linux distribution. Windows users are referred to msysgit. After git is installed, type:

git clone git://github.com/Ilastik/ilastik.git

to clone your own repository. Once you cloned the source code with git, you need to provide the following list of dependencies.

Dependencies

Ilastik has the following dependencies:

For most packages, it is convinient to download and install the binary build of the corresponding python package (e.g. PyQt). For linux also consider using easy_install.

Based on our experience we recommend the following steps if you encounter intallation problems.

Windows

qimage2ndarray

For Windows there is no precompiled binary of qimage2ndarray for Python 2.6 so far. To compile qimage2ndarray from sources, we recommend to use the mingw compiler compiler. First you need to download the corresponding Qt version. This choise depends on your PyQt4 installation. If you installed PyQt 4.7.3 you will have to download and install qt-win-opensource-4.6.3-mingw.exe from Qt 4.6.3. After you downloaded qimage2ndarray build and install it by:

python setup.py build --compiler=mingw32
python setup.py install

. If your Qt installation is not located in the default path (e.g. C:\Qt\4.6.3) you will need to edit the setup.py of qimage2ndarray and set the Qt include directories manually.

vigra

To install vigra and vigranumpy (its python bindings) follow the instactions on the vigra homepage. We recommend cmake >= 2.8 and boost-python libraries > 1.40. Binaries can be found on boostpro. Also a hdf5 > 1.8.0 is required.


Linux

numpy

We recommend to use the numpy version 1.5.0 for ilastik under linux. The easy_install command

easy_install -U numpy

will install and upgrade your numpy installation to version 1.5.0

Mac OS X

We are working on this, but constructive comments are welcome.