基于机器学习的视频表情识别研究

Research on video expression recognition based on machine learning

文章采用卷积神经网实时地进行视频中的人类表情识别。为了适应动态地进行表情识别的需要,文章专门设计了神经网系统,以满足实时地相应需求,并且利用视频中表情的变化更好地进行表情识别,文中的卷积神经网经过特别预训练和网络组合之后也提高了识别率。实验表明该系统能够实时在视频中识别表情。

In this paper, we proposed a new kind of framework of convolutional neural network, to recognize human facial expression. To adept the dynamic facial expression recognition, and responds to the need in real-time, after design neural network specially, this system improves the facial expression rate in the video flow, the experimental results show that the system can recognize facial expressions in video in real time.