Automatic facial action analysis

Abstract

This thesis describes a research system called aide: automated intelligent design expert that creates graphics from data. It assists users of data who wish to design graphics by combining artificial intelligence techniques with current automated design software paradigms. The work extends previous research in four ways. First, it combines the artificial intelligence technique known as case-based reasoning with traditional rule-based automated layout techniques. Case-based reasoning is used to choose a graphical template from a library of examples and is also used to instantiate the graphical aspects of the image. This approach is investigated because it is well matched to the process that people take when designing -- adapting old designs to new situations. Because of their cognitive importance, the rule-based components of aide have special design knowledge for temporal and locative information, allowing the system to present this type of information in unique ways. Like other automatic layout systems, aide is capable of producing traditional static graphics. In addition, however, aide can also design and present dynamic graphics. Finally, aide runs in a one-of-a-kind large high-resolution display environment. This, combined with a unique set of design skills and techniques, allows the system to produce graphical images of a higher quality than found previously. The research presented here contributes to the fields of artificial intelligence and automated graphic design by effectively combining them in one system. aide, the resulting prototype application, can be used to generate a number of different high-quality graphical representations from application independent data and offers potential benefits for anyone who needs to analyze or present data.