Synthetic Multimodal Dataset for Empowering Safety and Well-being in Home Environments

本数据集名为‘合成多模态家庭环境安全与福祉增强数据集’，由富士通有限公司和国家先进工业科学技术研究所联合创建。该数据集包含203个场景，总计1,218个视频和2,902,676个知识图谱三元组，旨在通过模拟日常活动视频和知识图谱，识别家庭环境中的危险情况。数据集创建过程中，使用了虚拟空间模拟器和知识图谱技术，自动生成图谱，无需手动标注。该数据集主要应用于家庭安全领域，特别是针对老年人的安全问题，通过分析人类行为和环境交互，预测并减少事故风险。

This dataset is named "Synthetic Multimodal Household Environmental Safety and Well-being Enhancement Dataset", which was jointly created by Fujitsu Limited and the National Institute of Advanced Industrial Science and Technology. It contains 203 scenarios, totaling 1,218 videos and 2,902,676 knowledge graph triples. The dataset aims to identify hazardous situations in household environments by simulating daily activity videos and knowledge graphs. During the creation process, virtual space simulators and knowledge graph technologies were adopted to automatically generate the graphs without manual annotation. This dataset is mainly applied in the field of household safety, especially for the safety issues of the elderly, by analyzing human behaviors and environmental interactions to predict and reduce accident risks.