文献类型：期刊文献

中文题名：基于深度学习的面部动作单元识别算法

英文题名：Facial Action Unit Recognition Algorithm Based on Deep Learning

作者：王德勋[1,2];虞慧群[1];范贵生[1]

机构：[1]华东理工大学计算机科学与工程系,上海200237;[2]上海市计算机软件测评重点实验室,上海201112

年份：2020

卷号：46

期号：2

起止页码：269

中文期刊名：华东理工大学学报（自然科学版）

外文期刊名：Journal of East China University of Science and Technology

收录：CSTPCD;;Scopus;北大核心:【北大核心2017】；CSCD:【CSCD_E2019_2020】；

语种：中文

中文关键词：面部动作单元识别;迁移学习;类别不平衡;动态加权损失;多任务训练

外文关键词：facial action unit recognition;transfer learning;unbalanced category;dynamic weighed loss;multitask training

摘要：面部动作单元识别任务是理解人脸表情最重要的环节之一,但因为类别极度不平衡和属于多标签分类等问题,给算法设计带来了不小的困难。针对这些问题设计了一种基于深度学习的面部动作单元识别算法。首先,基于迁移学习理论,以人脸识别任务为目标驱动,使用大规模数据集预训练卷积网络,使模型具有提取人脸抽象特征的能力;其次,设计了一个根据分类置信度来动态加权样本损失大小的目标函数,使得模型更关注于优化少数类样本;最后,结合多标签共现关系拟合和人脸关键点回归两个相关任务,联合训练模型并测试。实验结果表明,该方法在CK+和MMI数据集上能有效提升分类正确率与F1分数。
Facial action unit recognition task is one of the most important aspects of understanding facial expressions.However,the extremely unbalanced categories and the multi-label classification bring great difficulties for the design of recognition algorithm.By means of deep learning technique,this paper proposes a face action unit recognition algorithm.Firstly,based on the transfer learning theory and driven by the face recognition task,some largescale datasets are used to pre-train the convolutional network so as to make this model has the ability to extract abstract features of face.Secondly,an objective function is designed to dynamically weigh the sample loss according to the classification confidence for making the model focus more on optimizing a few samples.Finally,two related tasks including the multi-label co-occurrence relationship fitting and the key-point regression of face are combined to jointly train and test the model.It is shown from the experimental results that the proposed method can effectively improve the classification accuracy and F1-score on the relevant datasets CK+and MMI.