激光与光电子学进展, 2020, 57 (14): 141501, 网络出版: 2020-07-28
基于集成卷积神经网络的面部表情分类 下载: 1810次
Facial Expression Classification Based on Ensemble Convolutional Neural Network
图 & 表
图 7. 简单平均实验结果图。在(a) PublicTest和(b) PrivateTest下,EnsembleNet、ResNet-18、VGGNet-19GP的平均准确率随着epoch增加的变化对比情况
Fig. 7. Result graphs of simple average experiment. Comparison of the average accuracy of EnsembleNet, ResNet-18, and VGGNet-19GP under (a) PublicTest and (b) PrivateTest with the increase of epoch
图 8. 加权平均实验结果图。在(a) PublicTest和(b) PrivateTest下,EnsembleNet随着ResNet-18权重变化的波动情况,以及与ResNet-18、VGGNet-19GP的对比情况
Fig. 8. Result graphs of weighted average.Fluctuations of EnsembleNet with the change of ResNet-18 weights under (a) PublicTest and (b) PrivateTest, and the comparison with ResNet-18 and VGGNet-19GP
图 9. PublicTest上VGGNet-19GP、ResNet-18、EnsembleNet准确率变化
Fig. 9. VGGNet-19GP, ResNet-18, and EnsembleNet accuracy curves on PublicTest dataset
图 10. PrivateTest上VGGNet-19GP、ResNet-18、EnsembleNet准确率变化
Fig. 10. VGGNet-19GP, ResNet-18, and EnsembleNet accuracy curves on PrivateTest dataset
图 11. CK+数据集上VGGNet-19GP、ResNet-18、EnsembleNet准确率变化
Fig. 11. VGGNet-19GP, ResNet-18, and EnsembleNet accuracy curves on CK+ dataset
图 12. EnsembleNet在FER2013数据集上的混淆矩阵。(a) PublicTest Confusion Matrix;(b) PrivateTest Confusion Matrix
Fig. 12. Confusion matrix of EnsembleNet on the FER2013 dataset. (a) PublicTest Confusion Matrix; (b) PrivateTest Confusion Matrix
表 1FER2013和CK+数据集平均准确率
Table1. Average accuracy on the FER2013 and CK+ datasets%
|
表 2本文方法与现有方法在FER2013和CK+数据集上识别率对比
Table2. Comparison of proposed model with existing methods on the FER2013 and CK+datasets
|
周涛, 吕晓琪, 任国印, 谷宇, 张明, 李菁. 基于集成卷积神经网络的面部表情分类[J]. 激光与光电子学进展, 2020, 57(14): 141501. Tao Zhou, Xiaoqi Lü, Guoyin Ren, Yu Gu, Ming Zhang, Jing Li. Facial Expression Classification Based on Ensemble Convolutional Neural Network[J]. Laser & Optoelectronics Progress, 2020, 57(14): 141501.