UDIVA
收藏资源简介:
UDIVA数据集是由巴塞罗那大学和计算机视觉中心创建的一个大型多模态多视角数据集,专注于非剧本化的面对面双人互动。该数据集包含147名参与者在实验室环境中进行的188个互动会话,总时长90.5小时,使用多种视听和生理传感器记录。数据集不仅包括参与者的社会人口统计信息、自我和同伴报告的个性特征,还包括内部状态和关系概况。UDIVA数据集旨在通过提供丰富的上下文元数据,填补现有数据集在情境和参与者多样性方面的空白,从而推动从个体和双人角度对人类交流的全面分析。该数据集的应用领域包括早期诊断和干预、增强远程存在和个性化代理等,旨在解决人类交流中的复杂问题。
The UDIVA dataset is a large-scale multimodal, multi-view dataset created by the University of Barcelona and the Computer Vision Center, focusing on unscripted face-to-face dyadic interactions. It comprises 188 interaction sessions involving 147 participants conducted in a laboratory setting, with a total duration of 90.5 hours, recorded using a variety of audiovisual and physiological sensors. The dataset includes not only the participants' sociodemographic information, self-reported and peer-reported personality traits, but also their internal states and relationship profiles. The UDIVA dataset aims to fill the gaps in existing datasets regarding contextual diversity and participant variability by providing rich contextual metadata, thereby advancing comprehensive analyses of human communication from both individual and dyadic perspectives. Its application domains include early diagnosis and intervention, enhanced telepresence, personalized agents, and other related fields, with the goal of addressing complex challenges in human communication.



