基于人脸表情识别的课堂评价系统  

Classroom evaluation system based on face expression recognition

作　　者：马鹏涛 高钰皓 陈昱 Ma Pengtao;Gao Yuhao;Chen Yu(School of Computer and Control Engineering,Northeast Forestry University,Harbin 150040,China)

机构地区：[1]东北林业大学计算机与控制工程学院,哈尔滨150040

出　　处：《现代计算机》2024年第18期77-82,共6页Modern Computer

基　　金：东北林业大学大学生创新创业训练计划项目(202310225536)。

摘　　要：针对当下课堂评价具有主观性和延后性的问题,在完善人脸表情识别技术的基础上,探寻人脸表情与课堂评价之间的关系,构建出切实可行的课堂评价模型。借助神经网络软件PyTorch,将MobileNetV2与三种不同的神经网络模型在FER2013上进行测试以便对比分析。结果表明,综合准确率和模型大小,MobileNetV2在数据集表现最好。再对模型在MMAFEDB数据集训练后得到准确率更佳的模型后,在现存课堂评价体系上设置多种模式对应上课时不同状态。由此,提供一种新的课堂评价模型。Aiming at the current problem of subjectivity and delay in classroom evaluation,based on the improvement of face expression recognition technology,explore the relationship between face expression and classroom evaluation,and construct a practicable classroom evaluation model.With the help of neural network software PyTorch,MobileNetV2 is smoothly and tested with three different neural network models on FER2013 for comparative analysis.The results show that MobileNetV2 performs the best in the dataset in terms of combined accuracy and model size.After training the model on the MMAFEDB dataset to obtain a model with better accuracy,multiple modes were set up on the existing classroom evaluation system to correspond to different states during class.As a result,a new classroom evaluation model is provided.

关 键 词：人脸表情识别技术 MobileNetV2 课堂评价 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]