AVCAffe

我们介绍了AVCAffe，这是第一个由认知负荷和情感属性组成的视听数据集。我们通过在视频会议平台上模拟远程工作场景来记录AVCAffe，在该平台中，受试者协作以完成许多具有认知意义的任务。AVCAffe是最初收集的最大 (不是从互联网收集的) 英语情感数据集。我们招募来自18个不同国家的106名参与者，年龄范围为18至57岁，男女比例均衡。AVCAffe包含总共108小时的视频，相当于58,000多个剪辑以及基于任务的自我报告的地面真相标签，用于唤醒，效价和认知负荷属性，例如精神需求，时间需求，努力和其他一些。我们认为，鉴于对情感和认知负荷进行分类的固有困难，AVCAffe对于深度学习研究社区将是一个具有挑战性的基准。此外，我们的数据集通过促进创建学习系统以更好地自我管理远程工作会议，并进一步研究有关远程工作对认知负荷和情感状态的影响的假设，填补了现有的及时空白。

We introduce AVCAffe, the first audio-visual dataset incorporating both cognitive load and affective attributes. We collected AVCAffe by simulating remote work scenarios on video conferencing platforms, where participants collaborated to complete various cognitively meaningful tasks. AVCAffe is the largest English affective dataset collected from real-world settings (not scraped from the Internet). We recruited 106 participants from 18 distinct countries, aged between 18 and 57 years old, with a balanced gender ratio. AVCAffe contains a total of 108 hours of video content, equivalent to over 58,000 clips, along with task-based self-reported ground-truth labels for arousal, valence, and cognitive load attributes such as mental demand, temporal demand, effort, and several others. We argue that AVCAffe will serve as a challenging benchmark for the deep learning research community, given the inherent difficulties in classifying affective states and cognitive load. Furthermore, our dataset fills existing timely research gaps by facilitating the development of learning systems to better self-manage remote work meetings, and enabling further research on hypotheses regarding the impacts of remote work on cognitive load and affective states.

搜集汇总

背景与挑战

背景概述

AVCAffe是首个结合认知负荷和情感属性的视听数据集，包含108小时视频和58,000多个剪辑，覆盖18个国家106名参与者，提供丰富的自我报告标签，适用于深度学习研究和远程工作影响分析。