【摘要】 针对人脸表情识别精度不高和特征表征能力弱的问题,首先比较各种深度学习模型应用到表情特征提取的优缺点。其次在基于深度学习的表情识别模型构建方面进行研究并提出改进,通过将卷积神经网络作为表情特征提取结构,添加SVM作为顶层分类器,得到了一个兼顾表情识别能力和结构稀疏性的表情识别模型。最后与传统人脸表情识别算法进行对比实验,实验结果表明所提算法能够明显增强表情识别能力。更多还原

【Abstract】 This paper analyzes the advantages and disadvantages of various deep learning models applied to facial expression feature extraction.It studies and proposes an improved expression recognition model based on deep learning.By using the convolution neural network as the feature extraction structure and adding SVM as the top classifier,it obtain a facial expression recognition which can balance expression recognition and structural sparseness model.更多还原