Lifelog Access in VR via Filtering

Abstract

Archives of lifelog data are generally difficult to access and explore due to the large amount of data that they contain. Especially for visual data, that is, lifelog images, we suggest that the larger and more immersive screens of virtual reality headsets may provide a good way for efficient and pleasant lifelog exploration. Yet, standard interaction designs used for lifelog data on regular screens may not translate well into virtual reality. We propose a visualization-based approach where characteristics such as location, time, and type of content are visualized via maps, calendars, and tags, respectively. We expect that such a vision-based interface is easy to handle and allows people to explore the database and find information by filtering out images that fulfill certain criteria. Different implementations of these visualizations are evaluated in a comparative study testing both general exploration tasks and more targeted search tasks. Our results prove the feasibility of this idea, but also illustrate the relevance of how certain visualizations are implemented, and related functionality is provided. For example, while the map-based visualization was suffering from different shortcomings, participants were very positive about the tag-based part of the interface that allowed them to solve the required tasks efficiently. Our results provide a proof of concept for such a filtering- and visualization-based approach for lifelog access in virtual reality. The identified issues offer concrete suggestions for further development and necessary modifications.