Towards Improving the Understanding of Image Semantics by Gaze-based Tag-to-Region Assignments
- Eye-trackers have been used in the past to identify visual foci in images, find task-related image regions, or localize affective regions in images. However, they have not been used for identifying specific objects in images. In this paper, we investigate whether it is possible to assign image regions showing specific objects with tags describing these objects by analyzing the users' gaze paths. To this end, we have conducted an experiment with 20 subjects viewing 50 image-tag-pairs each. We have compared the tag-to-region assignments for nine existing and four new fixation measures. In addition, we have investigated the impact of extending region boundaries, weighting small image regions, and the number of subjects viewing the images. The paper shows that a tag-to-region assignment with an accuracy of 67% can be achieved by using gaze information. In addition, we show that multiple regions on the same image can be differentiated with an accuracy of 38%.
Author: | Tina Walber, Ansgar Scherp, Steffen Staab |
---|---|
URN: | urn:nbn:de:kola-5210 |
Series (Volume no.): | Arbeitsberichte, FB Informatik (2011,8) |
Document Type: | Part of Periodical |
Language: | English |
Date of completion: | 2011/05/31 |
Date of publication: | 2011/05/31 |
Publishing institution: | Universität Koblenz-Landau, Campus Koblenz, Universitätsbibliothek |
Release Date: | 2011/05/31 |
Tag: | gaze information; image semantics; tagging |
Number of pages: | 33 |
Institutes: | Fachbereich 4 / Institut für Informatik |
Dewey Decimal Classification: | 0 Informatik, Informationswissenschaft, allgemeine Werke / 00 Informatik, Wissen, Systeme / 004 Datenverarbeitung; Informatik |
Licence (German): | Es gilt das deutsche Urheberrecht: § 53 UrhG |