EMMI - Empathic Multimodal Motivational Interviews Dataset
收藏资源简介:
EMMI数据集是由ISIR Sorbonne大学巴黎创建,包含285个视频,总时长21小时22分钟,涵盖模拟的动机性访谈对话。数据集结合了AnnoMI和MID两个现有数据集,并增加了多模态注释,旨在分析治疗师和患者的互动行为,特别是情感和社交行为。创建过程中,数据集通过在线开放源收集,并进行了视频预处理和多模态注释,以支持虚拟治疗代理的开发。该数据集主要应用于心理治疗领域,特别是动机性访谈,以解决心理健康服务资源不足的问题,通过虚拟代理提供即时支持。
The EMMI Dataset was created by ISIR Sorbonne University Paris. It contains 285 videos with a total duration of 21 hours and 22 minutes, covering simulated motivational interviewing dialogues. This dataset integrates two existing datasets, AnnoMI and MID, and adds multimodal annotations, aiming to analyze the interactive behaviors between therapists and patients, especially their emotional and social behaviors. During its development, the dataset was collected from open online sources, followed by video preprocessing and multimodal annotation, to support the development of virtual therapy agents. This dataset is mainly applied in the field of psychotherapy, particularly motivational interviewing, to address the shortage of mental health service resources by providing real-time support via virtual agents.




