Interactive 2D and 3D Segmentation with ilastik
The ilastik system developed by our group uses machine learning and
simple interaction techniques to empower users without special image
processing expertise to segment and analyze their 2- and 3-dimensional
image data on their own. It offers a number of easy-to-use workflows
for various common analysis tasks. The talk will present two of these
workflows (“interactive classification” and “region carving”), going
from an online demonstration of the high-level user experience down to
the algorithmic and software design details. Special emphasis will be
put on aspects of genericity and parallelization which facilitate
convenient adaptation of basic building blocks to different contexts
without loss of performance. Examples from challenging biological
applications will illustrate our system's capabilities.
Ullrich Köthe is a senior researcher in the Multidimensional Image Processing Group at the Interdisciplinary Centre for Scientific Computing of the University of Heidelberg, Germany, where he heads advanced research on image and data analysis for industry and the life sciences. Formerly, he was with the Cognitive Systems Group at the University of Hamburg (1999-2007) and with the Fraunhofer-Institute for Computer Graphics in Rostock (1992-1999).
Ullrich Köthe's main field of research is the development of
user-friendly image analysis methods. He is interested in all
techniques that facilitate genericity and re-usability, from machine
learning and probabilistic modeling to generic programming and
empirical performance evaluation. Dr. Köthe is the originator and
maintainer of the open-source image analysis library VIGRA and
co-developer of the interactive ilastik toolkit, which have been
designed to put these ideas into practice.