|本期目录/Table of Contents|

[1]张东晓,陈彦翔.一种面向移动端的浅层CNN表情识别[J].集美大学学报(自然科学版),2021,26(2):129-138.
　ZHANG Dongxiao,CHEN Yanxiang.Mobile-Oriented Facial Expression Recognition Based on Shallow CNN[J].Journal of Jimei University,2021,26(2):129-138.
点击复制

一种面向移动端的浅层CNN表情识别(PDF)
分享到：

《集美大学学报（自然科学版）》[ISSN:1007-7405/CN:35-1186/N]

卷:
第26卷
期数:
2021年第2期
页码:
129-138
栏目:
数理科学与信息工程
出版日期:
2021-03-28

文章信息/Info

Title:
Mobile-Oriented Facial Expression Recognition Based on Shallow CNN
作者:
张东晓; 陈彦翔
(集美大学理学院,福建 厦门 361021)
Author(s):
ZHANG Dongxiao; CHEN Yanxiang
(School of Science，Jimei University，Xiamen 361021，China)
关键词:
面部表情识别; 卷积神经网络; 全局平均池化; Google Colab; Core ML
Keywords:
facial expression recognition; conrolutional neural networks(CNN); global average pooling; Google Colab; Core ML
分类号:
-
DOI:
-
文献标志码:
-
摘要:
移动端的表情识别有巨大需求，但是受算力限制，主流深度神经网络无法直接移植。为此，设计了一个浅层网络，在节约计算量的同时保证了识别率。网络中使用三组堆叠而成的卷积层，有助于增大感受野，便于更好地提取特征，这是提升识别率的关键；使用全局平均池化层，避免引入额外的全连接层，大幅降低参数量，在训练样本不足的情况下，降低模型过拟合风险。在FER-2013数据集进行训练，准确率超过现有大多数算法；在CK+数据集上进行微调，测试集上的准确率可达到0.96。将所得模型转换为Core-ML模型，结合Xcode平台在iOS端搭建了实时表情识别App，在iPhone 8 Plus上能够稳定、流畅运行，识别效果达到预期。
Abstract:
There is a huge demand for facial expression recognition on mobile terminals.However，due to the limitation of computational power，most popular deep neural networks cannot be directly transplanted.Therefore，a shallow network is designed in this paper，which can not only save the amount of calculation，but also ensure the recognition rate.The network uses three groups of stacked convolution layer，which helps to increase the receptive field and facilitate better feature extraction.This is the key to improve the recognition rate.The network also uses the global average pooling layer instead of additional full connection layer，which greatly reduces the number of parameters，and reduces the risk of overfitting in the case of insufficient training samples.The accuracy of model is higher than that of most existing algorithms on FER2013 data set.The accuracy rate can reach 0.96 by fine tuning on CK + data set.The model is transformed into the core ML model，and a realtime expression recognition app is built on the iOS side with Xcode platform.It can run stably and smoothly on the iPhone 8 plus，and the recognition effect reaches the expectation.

参考文献/References:

-

相似文献/References:

[1]曹长玉,郑佳春,黄一琦.基于区域卷积网络的行驶车辆检测算法[J].集美大学学报(自然科学版),2019,24(4):315.
　CAO Changyu,ZHENG Jiachun,HUANG Yiqi.Research on Running Vehicle Detection Algorithm-Based on Regional Convolution Network[J].Journal of Jimei University,2019,24(2):315.
[2]陈德意,张宏怡,刘彩玲,等.基于关键词策略和CNN的中文文本有害信息分类[J].集美大学学报(自然科学版),2020,25(5):392.
　CHEN Deyi,ZHANG Hongyi,LIU Cailing,et al.Classification of Chinese Text Harmful Information Based on Keywords Strategy and Convolutional Neural Network[J].Journal of Jimei University,2020,25(2):392.
[3]宋策,尹勇,王鹏.基于改进YOLOv5的船舶目标检测算法[J].集美大学学报(自然科学版),2023,28(2):136.
　SONG Ce,YIN Yong,WANG Peng.Ship Target Detection Algorithm Based on Improved YOLOv5[J].Journal of Jimei University,2023,28(2):136.
[4]花海波,于洪亮,闫锦,等.基于应力波卷积神经网络的齿轮故障诊断方法[J].集美大学学报(自然科学版),2023,28(4):335.
　HUA Haibo,YU Hongliang,YAN Jin,et al.Fault Diagnosis of Gear Based on Stress Wave Feature and Convolutional Neural Network[J].Journal of Jimei University,2023,28(2):335.
[5]蔡岱立,谢维波.FMSWFormer：基于频率分离和自适应多尺度窗口的视觉Transformer[J].集美大学学报(自然科学版),2023,28(6):568.
　CAI Daili,XIE Weibo.FMSWFormer:Visual Transformer with Frequency Separation and Adaptive Multi-Scale Window Attention[J].Journal of Jimei University,2023,28(2):568.

备注/Memo

备注/Memo:
-
更新日期/Last Update: 2021-05-17