人脸视频情感分类

VideoEmotionSet 是一个综合视觉与听觉信息的视频情感识别数据集，包含超过数十万段带有情绪标签的短视频片段。数据涵盖不同人种、性别、年龄、场景与语言背景，每段视频提供同步的人脸表情、语音语调与上下文信息，适用于多模态情绪识别、视频推荐系统、情感计算和人机交互等方向。该数据集支持对人类情绪在真实场景中的表达方式进行深入建模和分析。VideoEmotionSet is a multimodal video emotion recognition dataset integrating visual and auditory information, containing over hundreds of thousands of short video clips annotated with emotion labels. The dataset covers diverse groups including different ethnicities, genders, age ranges, scenarios, and language backgrounds. Each video provides synchronized facial expressions, speech intonation, and contextual information, making it applicable to research directions such as multimodal emotion recognition, video recommendation systems, affective computing, and human-computer interaction. This dataset enables in-depth modeling and analysis of human emotional expression in real-world scenarios.

搜集汇总

背景与挑战

背景概述

该数据集名为FaceEmotion-VideoSet，由Behavision发布，用于人脸视频情感分类任务，大小为36.29GB，采用Apache License 2.0许可证。数据集需登录后通过SDK或GIT方式下载，但未提供详细的介绍内容。