|
|
|
|
|
|
|
|
|
![]() |
|
![]() |
![]() |
![]() |
![]() |
||
|
Name:
Jari Kätsyri
Organization: Helsinki University of Technology |
Supervisor:
Professor Mikko Sams
Research Topic: Perceiving and modelling emotional expressions and non-verbal speech |
||||||
![]() |
|
![]() |
|||||
|
|
![]() |
|
|||||
![]() |
"Artificial Person" (AP) is a Talking Head, a three-dimensional computer animation model of a speaking person, developed at the Helsinki University of Technology. The development of AP is continuation for an earlier work on the same subject. In general, Talking Heads have been used e.g. as animated agents in user interfaces and as research stimuli in cognitive science experiments. An earlier Talking Head developed in HUT has also been used for such purposes. University of Tampere has applied the earlier Talking Head as a part of their museum guide interface. This Talking Head was also used in HUT in a research studying audiovisual speech perception. The research work being described aims at two things: 1) to research and evaluate facial expression and non-verbal behavior models for Talking Heads and 2) to apply the AP in cognitive science studies. When it comes to the first research goal, the most important part of the work is to create a reasonable, but computationally light facial expression model for the AP. The facial expressions will be based on a quite well-known psychological tool called Facial Action Coding System (FACS). FACS was developed from the basis of facial muscle anatomy to be a comprehensive tool capable of classifying all visually distinguishable actions on the face. FACS by itself isn't a tool for modelling facial expressions, though. FACS is meant to be used by a skilled (human) FACS coder for scoring videotaped facial behaviour. In addition, FACS describes facial actions on an objective level without considering e.g. affective interpretation of the actions. However, it is possible to create for example FACS-based "prototypes" for affective facial expressions. In fact, FACS "prototypes" of affective facial expressions have been used earlier as a basis for facial expressions in Talking Heads. The facial expression model of AP will be based on a rough approximation of the facial actions described by FACS system. Even if no explicit muscle model is used, this is a good basis for estimating real facial anatomy, since FACS was created from the basis of facial muscles. There are other implications as well. FACS has been used in studies considering nonverbal behaviour, such as facial movements relating to emphasizing vocal speech or behaviour indicating whose turn it is to speak in a conversation between two or more people ("turn-taking"). Using FACS gives a possibility to both apply results of these studies in the animation of AP and to use AP in similar studies. Talking Heads are very often able to express six basic expressions that are correlated to six emotions considered to be universal by some psychologists. Universal or not, at least these six basic expressions seem to be the most easily recognizable ones, so using them in expression models is understandable. However, six expressions is not much. In facial expression psychology there has been some research about blended expressions, where the blends contain elements from two basic expressions. Blended expressions have not been used in Talking Heads in before. A subgoal here is to create FACS "prototypes" for blended expressions and to implement them in the AP, thereby creating a larger repertoire of expressions. The second research goal was to apply the AP in actual cognitive science research. Most probably these studies will be related to (emotional and other behavioural) expression perception and processing in humans and they will cover both psychological and neurological experiments. |
||||||
![]() |
|||||||
![]() |
|
![]() |
|||||
|
|
|||||||
![]() |
Links: none |
![]() |
|
|
|||
![]() |
![]() |
||||||