The National New Energy and Industrial Technology Development Organization (NEDO) and Tokyo Denki University have converted video and audio such as human facial expressions during group communication and body movements detected by sensors into data for research institutes of universities and companies. Started offering.It is the world's first group communication dataset that can be used to design next-generation AI (artificial intelligence) systems, and is expected to accelerate the research and development of next-generation AI and robots.
According to Tokyo Denki University, the data includes 2 minutes of video, audio, and physical movements of discussions conducted by 6 second-year students and 6 working adults of Tokyo Denki University.The video was captured by a total of 100 cameras, the audio was captured by the headset worn by the participant, and the body movement was detected by sensors attached to the participant's head, chest, and both upper arms.
Behaviors such as remarks, gestures, and gaze are labeled so that the research institution that received the data can not only analyze the conversation but also use it as data for AI machine learning.
In order to advance the study of communication behavior, it is necessary to collect and weigh more data sets.For this reason, Tokyo Denki University has established a consortium for the purpose of collecting data of the same standard.
In addition to Tokyo University of Agriculture and Technology, researchers from more than 10 universities, including Tokyo University of Agriculture and Technology, Shonan Institute of Technology, and Toyohashi University of Technology, agree.Tokyo Denki University plans to invite other educational and research institutes to participate.