CREMA-D

该数据集名为CREMA-D，它是一个结合了面部和语音数据的音视频数据集，旨在对六种基本情绪状态进行分类。该数据集包含了91位演员在2至3秒的视频片段中说出几个短单词，共计7,442个片段。此外，该数据集涵盖了六种最常见的情绪。任务是对这些音视频数据进行视觉听觉情绪分类。

This dataset, named CREMA-D, is an audio-visual dataset integrating facial expression and speech data, designed for classifying six basic emotional states. It contains a total of 7,442 clips, where 91 actors spoke several short words within 2- to 3-second video segments. Additionally, this dataset covers the six most common emotional states. The corresponding task is to perform emotional classification by leveraging both visual and auditory information from these audio-visual data.

搜集汇总

背景与挑战

背景概述

CREMA-D是一个多模态情感数据集，包含7,442个来自91位不同背景演员的片段，涵盖六种情感和四种情感水平。数据集通过众包方式收集了2,443名参与者的评分，用于音频、视频和视听模态的情感识别研究。