No abstract

The costs of eye-tracking technologies steadily decrease. This allows research institutions to obtain multiple eye-tracking devices. Already, several multiple eye-tracker laboratories have been established. Researchers begin to recognize the subfield of group eye-tracking. In comparison to the single-participant eye-tracking, group eye-tracking brings new technical and methodological challenges. Solutions to these challenges are far from being established within the research community. In this paper, we present the Group Studies system, which manages the infrastructure of the group eye-tracking laboratory at the User Experience and Interaction Research Center (UXI) at the Slovak University of Technology in Bratislava. We discuss the functional and architectural characteristics of the system. Furthermore, we illustrate our infrastructure with one of our past studies. With this paper, we also publish the source code and the documentation of our system to be re-used.

No abstract

-Automatic image annotation methods require a quality training image dataset, from which annotations for target images are obtained. At present, the main problem with these methods is their low effectiveness and scalability if a large-scale training dataset is used. Current methods use only global image features for search. We proposed a method to obtain annotations for target images, which is based on a novel combination of local and global features during search stage. We are able to ensure the robustness and generalization needed by complex queries and significantly eliminate irrelevant results. In our method, in analogy with text documents, the global features represent words extracted from paragraphs of a document with the highest frequency of occurrence and the local features represent key words extracted from the entire document. We are able to identify objects directly in target images and for each obtained annotation we estimate the probability of its relevance. During search, we retrieve similar images containing the correct keywords for a given target image. For example, we prioritize images where extracted objects of interest from the target images are dominant as it is more likely that words associated with the images describe the objects. We tailored our method to use large-scale image training datasets and evaluated it with the Corel5K corpus which consists of 5000 images from 50 Corel Stock Photo CDs.