区域增强型注意力网络下的人脸表情识别  

Facial Expression Recognition Based on Region Enhanced Attention Network

作　　者：陈公冠 张帆 王桦 范辉[1] 张彩明 Chen Gongguan;Zhang Fan;Wang Hua;Fan Hui;Zhang Caiming(School of Computer Science and Technology,Shandong Technology and Business University,Yantai 264005;Shandong Future Intelligent Financial Engineering Laboratory,Yantai 264005;School of Information and Electrical Engineering,Ludong University,Yantai 264011;School of Software,Shandong University,Jinan 250101)

机构地区：[1]山东工商学院计算机科学与技术学院,烟台264005 [2]山东省未来智能金融工程实验室,烟台264005 [3]鲁东大学信息与电气工程学院,烟台264011 [4]山东大学软件学院,济南250101

出　　处：《计算机辅助设计与图形学学报》2024年第1期152-160,共9页Journal of Computer-Aided Design & Computer Graphics

基　　金：山东省高等学校青创科技计划(2019KJN042);国家自然科学基金(62007017,62072281,61902220)。

摘　　要：为了识别人脸表情中包含复杂背景、面部遮挡等因素的真实环境下的图像,提出基于区域增强型注意力网络的人脸表情识别方法.首先提出基于注意力的区域增强网络,减弱外部因素的影响以及增强表情识别在真实环境下的鲁棒性;然后提出通道-空间注意力融合网络,作用于全局的特征提取;最后通过分区损失和交叉熵损失相结合的方式提升表情图像的辨识度,从而提升识别准确率.在公开数据集RAF-DB,FERPlus和AffectNet上的实验结果表明,表情识别准确率分别达到88.81%,89.32%和60.45%;所提方法具有更高的准确率和鲁棒性.In order to recognize facial expression images in real environments including complex back-ground,facial occlusion and other factors,a facial expression recognition method based on region enhanced attention network is proposed.Firstly,an attention-based region enhancement network is proposed to reduce the influence of external factors and enhance the robustness of expression recognition in real environments.Then,a channel-spatial attention fusion network is proposed to extract global features.Finally,the recogni-tion degree of facial expression images is improved by the combination of partition loss and cross entropy loss,thereby improving the recognition accuracy.The experimental results on the public datasets RAF-DB,FERPlus and AffectNet show that their expression recognition accuracy is 88.81%,89.32%and 60.45%.In conclusion,the method in this paper has good accuracy and robustness.

关 键 词：人脸表情识别 区域增强 注意力融合 分区损失 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]