首页> 外文会议>International Conference on Automatic Face and Gesture Recognition >Multi View Facial Action Unit Detection based on CNN and BLSTM-RNN
【24h】

Multi View Facial Action Unit Detection based on CNN and BLSTM-RNN

机译:基于CNN和BLSTM-RNN的多视图面部动作单元检测

获取原文

摘要

This paper presents our work in the FG 2017 Facial Expression Recognition and Analysis challenge (FERA 2017) and we participate in the AU occurrence sub-challenge. Our work of AU occurrence recognition is based on deep learning, and we design convolution neural network (CNN) models for two types of work: facial view recognition and AU occurrence recognition. For facial view recognition, our model could achieve 97.7% accuracy on validation dataset about 9 facial views. For AU occurrence recognition, we use both visual features and temporal information of dataset. We use CNN models to get deep visual feature and then use BLSTM-RNN to learn the high-level feature in the time domain. When training models, we divide dataset into 9 parts based on 9 facial views, and each model is trained in a specific view. When recognizing AUs, we recognize facial view first and then choose the corresponding model for AU occurrence recognition. Finally, our method shows good performance, the F1 score of test data is 0.507 and the accuracy is 0.735.
机译:本文介绍了我们在FG 2017面部表情识别和分析挑战中的工作(FERA 2017),我们参加了AU发生的亚挑战。我们的AU发生识别的工作是基于深度学习,我们设计了两种类型的工作卷积神经网络(CNN)模型:面部视图识别和AU发生识别。对于面部视图认可,我们的模型可以在验证数据集中达到97.7%的准确性约为9个面部视图。对于AU发生识别,我们使用DataSet的视觉功能和时间信息。我们使用CNN模型来获得深度视觉功能,然后使用BLSTM-RNN在时域中学习高级功能。培训模型时,我们将数据集分为9个零件,基于9个面部视图,每个模型都在特定视图中培训。当识别AU时,我们首先识别面部视图,然后选择相应的AU发生识别模型。最后,我们的方法表现出良好的性能,测试数据的F1评分为0.507,准确度为0.735。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号