Deep learning framework for modeling the effect of VR spatial resolution in VR sickness assessment

Abstract

Recently, Virtual reality (VR) becomes more and more popular around the industry such as entertainment, simulation training, education and so on. Although VR contents can give immersive and dynamic experiences to viewers, there are many reports that VR sickness occurs during employing VR environments. VR sickness is one of the main issues which hamper development of VR industry. To resolve this VR sickness, researches for quantifying and assessing it are needed. In this paper, we propose a novel deep learning-based framework for modeling the effect that spatial resolution of VR contents causes VR sickness. The proposed method takes into account spatio-temporal visual perception of 360-degree video for assessing VR sickness score. To consider spatio-temporal visual perception of VR contents, we design deep model to predict not only VR sickness score but also spatial and temporal perception from the latent features. To evaluate the effectiveness of the proposed method for VR sickness assessment, we built a new dataset that consists of 360-degree videos (stimuli), physiological signals, and the corresponding subjective simulator sickness questionnaire (SSQ) scores by our subjective assessment experiment. Experimental results demonstrated that the proposed metric had a high correlation with human subjective score for VR sickness.