My research activities within the QGar group of the French LORIA laboratory mainly concern intelligent hyper-linked documents and the tools to automate their generation. The fundamental topics this addresses are image segmentation, object recognition, image indexing and data representation, as shown in the following talk on image semantics (in French) during the web semantics workshop organized by INRIA.
I'm currently involved with the Océ document processing company on this subject, and have several students working in this domain. A global description of our ongoing work can be found here.
I was formerly involved in the VIGOR Esprit-IV project, where I worked on visual robot control in industrial environments. More links and information can be found at my ancient homepage at the INRIA Rhône-Alpes' MOVI research group.