新型多模态个性识别数据集
收藏资源简介:
本研究构建了一个包含全身姿态数据的新型多模态个性识别数据集。该数据集由287名大学生完成的虚拟面试视频组成,视频包含36个问题,并伴有自我报告的大五人格评分作为标签。数据集采用了AlphaPose技术提取姿态信息,涵盖了视觉、音频、文本等多种模态,适用于心理学、教育、人工智能等领域,旨在提高个性预测模型的准确性。
This research develops a novel multimodal personality recognition dataset with full-body pose data included. The dataset is composed of virtual interview videos completed by 287 college students, which contain 36 questions, and is paired with self-reported Big Five Personality scores as ground-truth labels. AlphaPose is employed to extract pose information from the dataset, which covers multiple modalities including visual, audio and text. This dataset is applicable to disciplines such as psychology, education and artificial intelligence, and aims to enhance the accuracy of personality prediction models.




