1,003人情感视频数据

1,003人情感视频数据涵盖多人种、多种室内场景、多种年龄段、多种语言、多种情感，其中包括11种面部情感，15种内心情感。在标注方面，标注每段视频中每句话的情感种类（包括面部情感和内心情感）、起止时间、文本转写。1,003人情感视频数据可用于情感识别、情感分析等任务，为语音识别相关研究及应用提供了丰富的资源，经多家AI公司验证：有助于模型面对真实世界的多样性时能够表现出色。我们严格遵循数据保护法规和隐私规定，确保数据采集、存储和使用的过程中维护用户的隐私和合法权益，所有数据均遵循GDPR, CCPA, PIPL。

The 1,003-person emotional video dataset covers multiple ethnic groups, diverse indoor scenes, various age groups, multiple languages and a wide range of emotions, including 11 types of facial emotions and 15 types of inner emotions. For annotation, the emotion category (including both facial and inner emotions), start and end timestamps, and text transcription of each sentence in every video are labeled. This dataset can be applied to tasks such as emotion recognition and sentiment analysis, providing abundant resources for research and applications related to speech recognition. Verified by multiple AI companies, it helps models perform excellently when facing the diversity of the real world. We strictly comply with data protection regulations and privacy rules, ensuring that user privacy and legitimate rights and interests are protected during the collection, storage and use of the data. All data adheres to GDPR, CCPA and PIPL.

搜集汇总

背景与挑战

背景概述

该数据集包含1003人的情感视频，涵盖多种族、室内场景、不同年龄段和语言，标注了11种面部情感和15种内心情感，包括情感类型、时间戳和文本转录，适用于情感识别与分析任务。数据以.mp4视频和.json标注文件格式提供，采集于室内环境，并严格遵守数据隐私法规。