查询结果:   马中启,朱好生,杨海仕,王琪,胡燕海.基于多特征融合密集残差CNN的人脸表情识别[J].计算机应用与软件,2019,36(7):197 - 201.
中文标题
基于多特征融合密集残差CNN的人脸表情识别
发表栏目
人工智能与识别
摘要点击数
445
英文标题
FACIAL EXPRESSION RECOGNITION BASED ON MULTI-FEATURE FUSION DENSE RESIDUAL CNN
作 者
马中启 朱好生 杨海仕 王琪 胡燕海 Ma Zhongqi Zhu Haosheng Yang Haishi Wang Qi Hu Yanhai
作者单位
宁波大学机械工程与力学学院 浙江 宁波 315211 宁波戴维医疗器械股份有限公司 浙江 宁波 315712    
英文单位
College of Mechanical Engineering and Mechanics, Ningbo University, Ningbo 315211, Zhejiang, China Ningbo David Medical Device Co., Ltd., Ningbo 315712, Zhejiang, China    
关键词
表情识别 密集型卷积神经网络 特征融合 深度学习
Keywords
Expression recognition Dense residual CNN Feature fusion Deep learning
基金项目
国家自然科学基金项目(51705263);宁波市重大科技专项(2017C110030)
作者资料
马中启,硕士生,主研领域:计算机视觉,深度学习。朱好生,工程师。杨海仕,工程师。王琪,助工。胡燕海,教授。 。
文章摘要
传统人脸表情识别主要基于人工提取特征,其存在算法鲁棒性较差、易受人脸身份信息干扰等问题,以及传统卷积神经网络不能充分提取人脸表情特征的现状。对此提出一种基于多特征融合密集残差卷积神经网络的人脸表情识别。该方法能够充分利用神经网络中每层的特征,在密集块中,对于每一个卷积层,其前面所有卷积层的输出都将作为本卷积层的输入。然后将每个密集块的输出送入到全连接层中进行特征融合,经过Softmax分类器分类。在CK+和FER2013数据集上进行多次实验,与传统的机器学习方法相比,该方法具有较高的准确率与较强的鲁棒性。
Abstract
Because the traditional facial expression recognition is mainly based on the artificial extraction of features, the robustness of the algorithm is poor, and it is easy to be interfered by the face identity information. Traditional CNN cannot adequately extract facial expression features. We proposed a facial expression recognition based on multi-feature fusion dense residual CNN. This method could make full use of the characteristics of each layer in the neural network. For each convolution layer in the dense block, the output of all convolution layers in front of it would be the input of this convolution layer. Then the output of each dense block was fed into the full connection layer for feature fusion, and classified by Softmax classifier. Experiments were performed on CK+ and FER2013 data sets. Compared with traditional machine learning, our method has higher accuracy and stronger robustness.
下载PDF全文