Sie sind hier


The K-Space Annotation Tool (KAT) is a framework for semi-automatic and efficient annotation of multimedia content. The KAT is developed within the K-Space Network of Excellence, and is released as open source under the terms of the LGPL.

The KAT provides the following features:

  • GUI Framework
  • Plugin Infrastructure
  • Formal model based on the Core Ontology on Multimedia (COMM)
  • A set of standard plugins for image annotation, retrieval, and browsing of ontologies

The vision behind the KAT is to integrate automatic and manual annotation of images, in order to improve the efficiency of the annotation process for a user. Automatic annotation plugins might provide segmentations of images or videos, might organize content collections using clustering techniques or might propose semantic annotations. A user can inspect these results and refine or correct them if required. Further, we also anticipate the development of interactive approaches, where the user gives input, in order to control the algorithm. An example is interactive segmentation, where a user might indicate the regions he is interested in, and the ones he is not interested in. Using this information, an algorithm might better be able to provide a correct segmentation.

Please visit the following pages to learn more about the KAT:

If you have any questions, please contact Carsten Saathoff.