用于视频情感识别的时空网络
计算机视觉与模式识别
2017-04-13 v3
摘要
我们的实验将几种流行的深度学习方法以及一些传统方法应用于视频情感识别问题。在实验中，我们使用 CNN-LSTM 架构进行视觉信息提取与分类，并利用传统方法进行音频特征分类。对于多模态融合，我们使用传统的支持向量机。我们的实验在 AFEW 6.0 数据集上取得了良好的结果。
引用
@article{arxiv.1704.00570,
title = {Spatiotemporal Networks for Video Emotion Recognition},
author = {Lijie Fan and Yunjie Ke},
journal= {arXiv preprint arXiv:1704.00570},
year = {2017}
}
备注
The reason is that the article is just an experimental report without being reviewed carefully. There are some fatal drawbacks in this article and it may not be suitable for being published