To cope with the large bandwidth and low-latency requirements, Virtual Reality (VR) systems are steering toward user-centric sys- tems in which coding, streaming, and possibly rendering are person- alized to the final user. The success of these user-centric VR systems mainly relies on the ability to anticipate viewers navigation. This has motivated a large attention in studying the prediction of user’s movements in a VR experience. However, most of these work lack of a proper and exhaustive behavioural analysis in a VR scenario, leaving many key-behavioural questions unsolved and unexplored: Can some users be more predictable than others? Do users have their own way of navigating and how much is this affected by the video content features? Can we quantify the similarity of users navigation? Answering these questions is a crucial step toward the understanding of user’s behaviour in VR; and it is the overall goal of this paper. By studying VR trajectories across different contents and through information-theoretic tools, we aim at characterizing navigation patterns both for each single viewer (profiling individ- ually viewers – intra-user analysis) and for a multitude of viewers (identifying common patterns among viewers – inter-user analysis). For each of these proposed behavioural analyses, we describe the applied metrics and key observations that can be extrapolated.
The paper will be presented at MMVE’20, June 8, 2020.
Please cite our paper in your publications if it helps your research: