电子学报

• 学术论文 • 上一篇    下一篇

基于兴趣区域深度神经网络的静态面部表情识别

孙晓, 潘汀   

  1. 合肥工业大学计算机与信息学院, 安徽合肥 230009
  • 收稿日期:2016-02-16 修回日期:2016-06-14 出版日期:2017-05-25
    • 作者简介:
    • 孙晓 男,1980年出生,山东龙口人,博士,副教授,硕士生导师.研究方向为人机交互系统、情感计算与机器学习.E-mail:sunx@hfut.edu.cn;潘汀 男,1995年出生,江苏连云港人,本科生.研究方向为深度学习,贝叶斯学习理论及其在计算机视觉与自然语言处理方面的应用.E-mail:neopenx@mail.hfut.edu.cn
    • 基金资助:
    • 安徽省自然科学基金 (No.1508085QF119); 国家自然科学基金重点项目 (No.61432004); 模式识别国家重点实验室开放课题 (No.NLPR201407345); 中国博士后科学基金 (No.2015M580532)

Static Facial Expression Recognition System Using ROI Deep Neural Networks

SUN Xiao, PAN Ting   

  1. School of Computer and Information, Hefei University of Technology. Hefei, Anhui 230009, China
  • Received:2016-02-16 Revised:2016-06-14 Online:2017-05-25 Published:2017-05-25
    • Supported by:
    • National Natural Science Foundation of Anhui Province,  China (No.1508085QF119); Key Program of National Natural Science Foundation of China (No.61432004); Open Project of National Laboratory of Pattern Recognition (No.NLPR201407345); China Postdoctoral Science Foundation (No.2015M580532)

摘要:

通过在面部表情数据集上训练深度卷积神经网络、深度稀疏校正神经网络两种模型,对两种深度神经网络在静态面部表情识别方面的应用作了对比和分析.基于面部表情的结构先验知识,提出一种面向面部表情识别的改良方法——K兴趣区域方法,该方法在构建的开放实验数据集上,降低了由于训练数据过少而导致深度神经网络模型泛化能力不佳的问题,使得混合模型普遍且显著地降低了测试错误率.进而,结合实验结果进行了深入分析,并对深度神经网络在任意图像数据集上的可能有效性进行了深入剖析和分析.

关键词: K兴趣区域, 深度神经网络, 深度学习, 面部表情识别

Abstract:

By building two models including Deep Convolutional Neural Networks and Deep Sparse Rectifier Neural Networks on facial expression dataset,we made contrastive evaluations in facial expression recognition system with deep neural networks.Based on prior structure knowledge of facial expression,we proposed a fast and simple improved method called K Region Of Interest——‘K-ROI’,which relieved the poor generalization of deep neural networks on experimental dataset due to insufficient data and decreased the testing error rate apparently and generally.Finally,we infer the experimental results and analyze comprehensively for the possible validity with deep neural networks on arbitrary image dataset.

Key words: K-ROI, deep neural networks, deep learning, facial expression recognition

中图分类号: