Read My Mind
收藏arXiv2023-03-07 更新2024-08-06 收录
下载链接：
http://arxiv.org/abs/2304.14501v1
下载链接
链接失效反馈官方服务：
资源简介：
Read My Mind是一个大规模的多模态视频数据集，旨在通过对象-上下文关系预测人类信念，以促进人机交互领域的发展。该数据集包含900个视频，总计347,490帧，涉及15个不同的上下文，通过非言语沟通信号如眼神、手势、姿态等捕捉人类行为。数据集的创建涉及使用多种技术如对象检测、姿态估计和眼神追踪，以精确标注参与者的隐含信念。此数据集的应用领域主要集中在理解和预测人类信念，以优化人机协作环境中的互动效率。
Read My Mind is a large-scale multimodal video dataset that predicts human beliefs via object-context relationships, with the goal of advancing the field of human-computer interaction. Comprising 900 videos totaling 347,490 frames, the dataset covers 15 distinct contexts and captures human behaviors through nonverbal communication cues such as eye gaze, gestures, and postures. The construction of this dataset utilizes multiple technologies including object detection, pose estimation, and eye tracking to accurately annotate the implicit beliefs of participating subjects. The primary application scenarios of this dataset focus on understanding and predicting human beliefs, so as to optimize interaction efficiency in human-computer collaboration environments.
提供机构：
新加坡国立大学
创建时间：
2023-03-07
