The research focuses on understanding human emotions, particularly fear, in virtual reality (VR) environments. This understanding is crucial in the development of immersive games, scene development, and next-generation human-computer interaction applications. The researchers used VR horror games to analyze fear emotions, collecting multi-modal data from 23 players. They used a model to predict fear with varying accuracies depending on the classification level. The resulting dataset, VRMN-bD, is compared favorably with existing datasets, showcasing fewer limitations in collection method, data scale, and audience scope.
Publication date: 25 Jan 2024
Project Page: https://github.com/KindOPSTAR/VRMN-bD
Paper: https://arxiv.org/pdf/2401.12133
