Being able to understand and predict users’ navigating pattern has a vital role in VR adaptive streaming. Namely, it would allow to transmit selective content based on the final users’ interest. On this topic, most of the recent works in the literature analyse omnidirectional navigation patterns in the planar domain: each video frame is projected into a plane by applying a map projection. The planar analysis allows for an easy visualization and qualitative assessment of users’ behaviour, but cannot be used for quantitative assessment since it neglects the actual content geometry (the physical spherical distances are not preserved). To develop sphere-based learning methods to predict users’ movements, we propose to analyse users’ behaviour directly on the sphere with extending machine learning algorithms to the spherical domain.
To know more about our research in this area:
- Spherical Clustering of users navigating in VR content
- Do Users Behave Similarly in VR? Investigation of the Influence on the System Design
- Understanding User Navigation in Immersive Experience: an Information-Theoretic analysis
- Influence of Narrative Elements on User Behaviour in Photorealistic Social VR