Overview

In our study, we needed a video database with different frame resolutions to conduct objective assessment of VR contents. For this reason, we created a new database of 360-degree videos for our validation experiments, that consist of 80 videos. The videos are collected from Vimeo and YouTube, all having 30 FPS. 17 subjects had participated in the data collection, and we collected individual subjective SSQ score for each participant.

This web page provides the datasets of our conference paper at 2019 IEEE International Conference on Image Processing (ICIP) [1], so that researchers can repeat our experiments on our datasets. The dataset is for research purposes only. If you use our datasets, please cite the paper [1].

360-degree Video Dataset

We collected 20 different 360-degree videos that include various scenes from Vimeo and Blend as a benchmark. The collected videos are preprocessed so that each video can be displayed with four different frame resolutions (SD, HD, FHD, UHD). The dataset consists of 80 videos in total. We conducted extensive subjective experiments, and each test video was presented for 60 seconds.

Video datasets :  [1]   [2]   [3]   [4]   [5]   [6]   [7]   [8]   [9]   [10]   [11]   [12]   [13]   [14]   [15]   [16]   [17]   [18]   [19]   [20]

Time Stamp Information for the Video dataset : [Link]

Subjective SSQ Score

In subjective assessment experiments, Oculus Rift CV1 was used for displaying 360-degree videos, which was one of the high-end stereoscopic type HMDs. Its display resolution is 2160 × 1200 pixels (1080 × 1200 pixels per eye). Its display frame rate is maximum 90 Hz and it has 110 degree FoV. A total of 17 subjects, aged 20 to 30, participated in our subjective experiments under the approval of KAIST Institutional Review Board (IRB). In general, the use of VR is not recommended for young people under 12 years of age due to immature development of visual-vestibular sensors. The participants in our experiment do not have health problems such as immature development of visual-vestibular sensors, vestibular dysfunction or oculomotor dysfunction, compared to children and older people. Subjects have normal or corrected-to-normal vision and minimum stereopsis of 60 arcsec. In our experiment, before watching each stimulus, they were placed in the center position to be started from zero position in order to prevent significantly different viewing traces between viewers They were seated on a rotatable chair in order to freely look around 360-degree contents. A week before the actual subjective assessment experiments, we had subjects experience a variety of VR contents with Oculus Rift in order to allow them to familiarize with VR environment. In our experiments, the subject head motion was small and negligible during watching 360-degree contents. Since most of the 360 degree-videos used in our experiment have movement in a certain direction by roller coaster and car, subjects focused their gaze in the similar direction (e.g., the direction of rails in the roller coaster video or moving direction in the driving video). The head motion below the range of 44º to 55º in yaw could not cause severe VR sickness. All experimental environments followed the guideline as per the recommendations of ITU-R BT.500-13 and BT.2021.

Subjective SSQ score datasets : [Link]

Reference

If you use the database, please cite as :

[1] K. Kim, S. Lee, H. G. Kim, M. Park and Y. M. Ro, "Deep Objective Assessment Model Based on Spatio-Temporal Perception of 360-Degree Video for VR Sickness Prediction," 2019 IEEE International Conference on Image Processing (ICIP), Taipei, Taiwan, 2019, pp. 3192-3196.