Semantic analysis of human visual attention in mobile eye tracking applications

This paper presents semantic analysis in mobile eye tracking studies as a novel application for video analysis in usability research and marketing analyses. Image analysis in the context of human attention within natural environments is recognized as a specific challenge on object detection and localization. In a characteristic study setting of urban marketing analysis, we propose an innovative method to extract semantic information from public media screens mediating content to passengers in public transportation. In an extensive performance test within a unique 100 subject study on real world data the method performed very successful in robustness and accuracy with respect to detection and thus proved being a valuable tool for semi-automated annotation of mobile eye tracking studies, requiring manual interaction in only ≈14% of image data.