PUB - Publikationen an der Universität Bielefeld

Human hand gestures are very swift and difficult to observe from the (often) distant perspective of a scientific overhearer. Not uncommonly, fingers are occluded by other body parts or context objects and the true hand posture is often only revealed to the addressee. In addition to that, as the hand has many degrees of freedom and the annotation has to cover positions and orientations in a 3D world – which is less accessible from the typical computer-desktop workplace of an annotator than, let’s say, spoken language – the annotation of hand postures is quite expensive and complex. Fortunately, the research on virtual reality technology has brought about data gloves in the first place, which were meant as an interaction device allowing humans to manipulate entities in a virtual world. Since its release, however, many different applications have been found. Data gloves are devices that track most of the joints of the human hand and generate data-sets describing the posture of the hand several times a second. The article reviews different types of data gloves, discusses representation formats and ways to support annotation, and presents best practices for study design using data gloves as recording devices.