基于C3D CNN的人脸表情识别系统设计与开发

Design and Development of Facial Expression Recognition System Based on Deep Learning

为了实现对人脸表情的自动识别,笔者设计和开发了一款基于C3D卷积神经网络(Convolutional Neural Network,CNN)的人脸表情识别系统。首先,利用已有Cohn-Kanade数据集和CASMEⅡ数据集作为训练数据。其次,使用Keras和TensorFlow的深度学习框架搭建C3D CNN,创建数据集并进行训练,以得到人脸表情识别模型。最后,使用PyQt5设计和开发人脸表情识别系统。结果表明,该系统具有页面简洁明了、方便用户操作等特点,可为心理诊断等领域提供一定的判断依据。

In order to realize the automatic recognition of facial expression, the author designs and develops a facial expression recognition system based on C3D Convolutional Neural Network(CNN). Firstly, the existing Cohn-Kanade data set and CASMEⅡ data set are used as training data. Secondly, the C3D CNN is built using the deep learning framework of keras and TensorFlow, and the data set is created for training to obtain the facial expression recognition model. Finally, we use PyQt5 to design and develop a facial expression recognition system. The results show that the system has the characteristics of simple and clear pages, convenient for users to operate and so on. Automatic recognition of facial expressions can provide a certain basis for psychological diagnosis, criminal interrogation and other fields.