基于注意力改进MobileNet-V3的教育机器人手势互动研究  

Research on Educational Robot Gesture Interaction Based on Attention Improvement of MobileNet-V3

作　　者：李娜[1] 余茜[1] LI Na;YU Xi(Xi’an siyuan University,Xi’an 710038,China)

机构地区：[1]西安思源学院,西安710038

出　　处：《自动化与仪器仪表》2025年第1期295-299,共5页Automation & Instrumentation

摘　　要：研究旨在技术创新提升大学生教育机器人在手势识别方面的性能,进而改善教育体验和教学效果。因此,研究提出了一种结合时间与空间的注意力机制对3D化改造后的MobileNet-V3模型进行进一步改进。实验结果显示,当在Jester数据集中超参数取值0.03,Hand数据集中超参数取值0.02后,两个数据集上的准确率均随着迭代次数增加而增加,LOSS值均随着迭代次数增加而减少。研究提出的与主流手势识别模型相比,在两个数据集上的平均识别准确率分别达到98.31%与98.48%,均优于对比模型。研究结论证实了注意力改进MobileNet-V3模型在教育机器人手势互动识别中的有效性,并展示了其在实际大学生教育环境中的应用潜力。The aim of this study is to improve the performance of university student educational robot in gesture recognition through technological innovation,so as to improve the educational experience and teaching effect.Therefore,an attentional mechanism combining time and space is proposed to further improve the 3D modified MobileNet-V3 model.The experimental results show that compared with the mainstream gesture recognition model,the average recognition accuracy of the proposed gesture recognition model on the two data sets is 98.31%and 98.48%,respectively,which is superior to the comparison model.In addition,the robots configured with the research model scored as high as 9.65 points in overall satisfaction during a one-month trial period at the university.The results confirm the effectiveness of the attention-improved MobileNet-V3 model in gesture recognition of educational robots,and demonstrate its potential application in the actual education environment of college students.

关 键 词：教育 机器人 注意力 手势识别 互动 

分 类 号：TP242[自动化与计算机技术—检测技术与自动化装置]