In this thesis we investigate visual place categorization by combining successful global image descriptors with a method of visual attention in order to automatically detect meaningful objects for places. The idea behind ...