Gudula Retz-Schmidt's research while affiliated with University of Hamburg and other places
What is this page?
This page lists the scientific contributions of an author, who either does not have a ResearchGate profile, or has not yet added these contributions to their profile.
It was automatically created by ResearchGate to create a record of this author's body of work. We create such pages to advance our goal of creating and maintaining the most comprehensive scientific repository possible. In doing so, we process publicly available (personal) data relating to the author as a member of the scientific community.
If you're a ResearchGate member, you can follow this page to keep up with this author's work.
If you are this author, and you don't want us to display this page anymore, please let us know.
It was automatically created by ResearchGate to create a record of this author's body of work. We create such pages to advance our goal of creating and maintaining the most comprehensive scientific repository possible. In doing so, we process publicly available (personal) data relating to the author as a member of the scientific community.
If you're a ResearchGate member, you can follow this page to keep up with this author's work.
If you are this author, and you don't want us to display this page anymore, please let us know.
Publications (9)
Die vorliegende Arbeit behandelt die Interpretation des Verhaltens mehrerer Akteure in Szenenfolgen. Es wurde ein planerkennungsbasierter Ansatz hierfür vorgestellt, der Verfahren zur Erkennung und Beschreibung von Intentionen, Interaktionen und Fehlschlagsursachen integriert. Ein System, das über solche Verfahren verfügt, ist in der Lage, intentio...
In diesem Abschnitt wird die in der vorliegenden Arbeit verwendete Terminologie eingeführt. Sie orientiert sich vor allem an der Terminologie der philosophischen Handlungstheorie (vgl. Abschnitt 2.1.2) und an der Terminologie der KI-Gebiete Plangenerierung (vgl. Abschnitt 2.3.1) und Planerkennung (vgl. Abschnitt 2.3.3).
Die Planerkennung ist von zentraler Bedeutung für die vorliegende Arbeit. Neben der Intentionsextraktion setzen die Erkennung von Interaktionen und die Analyse von Fehlschlagsursachen darauf auf. Alle drei auf der Planerkennung aufsetzenden Prozesse geben ihre Ergebnisse an die Beschreibungskomponente weiter, die die übermittelte Information in ein...
Die vorliegende Arbeit ist im Umfeld der natürlichsprachlichen Beschreibung von Bildfolgen entstanden, einem Teilbereich des Aufgabengebietes „Kopplung von bildverstehenden und sprachverstehenden Systemen“. In diesem Bereich hat man sich bisher nur mit der Erkennung und Beschreibung von Objekten sowie deren räumlichen Relationen und Bewegungen, als...
Die im Rahmen der vorliegenden Arbeit entwickelten Ansätze zur automatischen Erkennung von Plänen und Intentionen wurden implementiert in dem System REPLAI-II. Die Beschreibung orientiert sich daher in diesem und im folgenden Kapitel an den Komponenten des Systems.
Natural language systems for the description of image sequences have been developed (e.g. Neumann and Novak, 1986; Herzog et al., 1989). Even though these systems were used to verbalize the behaviour of human agents, they were limited in that they could only describe the purely visual, i.e. spatio-temporal, properties of the behaviour observed. For...
The system REPLAI-II extends the range of capabilities of natural language systems describing real-world image sequences. The systems developed so far in this field merely extract spatio-temporal information from image sequences and verbalize it in natural language. Such a system is, for instance, the system SOCCER, which describes extracts of socc...
For the recognition and verbalization of events in real-world image sequences it is in certain cases necessary to have expectations, which are based on common-sense knowledge. This paper deals with the use of scripts as a means of generating expectations about events in the domain of traffic scenes and with the evaluation of these expectations for...
SFB 314, Project VITRA, Universität des Saarlandes D-66041 Saarbrücken Zusammenfassung Im Hinblick auf die Integration von maschinellem Sehen und der Verarbei-tung natürlicher Sprache setzt sich der vorliegende Beitrag genauer mit dem Pro-blem der simultanen Analyse und natürlichsprachlichen Beschreibung einer Re-alweltbildfolge auseinander. Als Di...
Citations
... Events often describe derived state changes on the physical level. An example for a system performing event recognition is the automated soccer commentator system SOCCER [HRS89], which analyzes physical trajectories of moving objects from sequences of real-world image data to extract events like running or passing the ball. Event recognition ranges from simple approaches like thresholding on a single sensed variable to complex approach that for instance consider the evolution of large and/or fused sets of observed variables over longer periods of time [WKY + 09]. ...
... scripts, cf. Retz-Schmidt [1985]) are required. ...
... These systems can verbalize the behaviour of human agents in image sequences about football and describe the spatio-temporal properties of the behaviour observed. Retz-Schmidt (1991) and Retz-Schmidt and Tetzlaff (1991) describe an approach which yields plan hypotheses about intentional entities from spatio-temporal information about agents. The results can be verbalized in natural language. ...
... Once the ground facts have been updated we compile derived facts F infer which are based on tracking sets of ground facts (e.g. the ball-player distances for ball ownership). For an efficient qualitative mapping we implemented a domain-specific focus heuristic based on suggestions by Retz-Schmidt et al. [12] that filters the available quantitative input in a preprocessing step such that only relevant scene aspects are considered. Although the focus is in principle suited for multiple key objects, we were concerned only with the detection of ball-centered motion situations. ...
