融合注意力机制的人机交互儿童情感识别技术研究  

Research on Human-Computer Interactive Children’s Emotional Recognition Technology Integrating Attention Mechanism

作　　者：朱芳慧[1] ZHU Fanghui(Xi’an Siyuan University,Xi’an 710038)

机构地区：[1]西安思源学院,西安710038

出　　处：《自动化与仪器仪表》2024年第7期251-255,共5页Automation & Instrumentation

基　　金：2021年西安思源学院校级精品在线开放课程“中外教育思想史”建设成果;2023年西安思源学院课程思政示范课程《中外教育史》及教学团队建设成果;西安思源学院科研创新团队“陕西区域基础教育科研创新团队”建设成果;西安思源学院校级教学团队“教育学核心课程群教学团队”建设成果。

摘　　要：随着人工智能的发展,越来越多的智能设备被应用于学前教育儿童的教育中。尽管情感识别技术在许多方面取得了显著的进展,但仍然存在一些挑战。其中之一是如何更好地适应于特定用户群体,尤其是对于儿童群体的情感识别。儿童的情感表达方式可能更为复杂,因此,需要更为细致入微的技术手段来进行有效的情感识别。研究通过梅尔语谱图和文本信息的词嵌入向量构建一个双模态的情感识别模型。实验结果表明,当数据集大小为500时,所提出的算法模型的召回率是0.91,F1值是0.92。损失函数值为0.12,准确率为0.91。研究结果表明,所提出的方法在儿童情感识别中有着较好的性能。With the development of artificial intelligence,more and more intelligent devices are being applied to the education of preschool children.Although emotion recognition technology has made significant progress in many aspects,there are still some challenges.One of them is how to better adapt to specific user groups,especially for emotional recognition of children.Children's emotional expression may be more complex,so more meticulous technical means are needed for effective emotional recognition.This study constructs a bimodal emotion recognition model using Mel spectrograms and word embedding vectors of text information.The experimental results show that when the dataset size is 500,the proposed algorithm model has a recall rate of 0.91 and an F1 value of 0.92.The loss function value is 0.12,with an accuracy of 0.91.The research results indicate that the proposed method has good performance in children’s emotional recognition.

关 键 词：人机交互 学前教育 情感识别 MEL频率倒谱系数 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]