基于深度学习的人脸表情识别系统.docx

PAGE 29 基于深度学习的人脸表情识别系统 摘 要 随着社会的进步和经济的发展，人工智能已经开始应用于各种各样的场景，最典型的应用就是机器人的应用。人机交互的设计已经越来越成熟，而机器人要想了解人的正确想法就不应仅体现在语言上，还应该在其他方面分析出人的正确情感，表情识别分析就是一个主要攻克点。本文从自然神经网络的介绍一直到运用卷积神经网络搭建模型进行具体的分析，包括自然神经网络和卷积神经网络的知识归纳，包括一些数学推导过程，对模型搭建过程和结果分析，会将所用的知识点进行详细归纳让人更好地理解和实现实际的模型搭建。本文的模型搭建最终使用2012年ILSVRC的冠军AlexNet实现，且会对模型进行分析。结尾也会适当提出运用VGG网络、ResNet、GoogLeNet等具有较多层神经网络的方法对模型的改进。 关键词：表情识别、卷积神经网络、知识归纳、数学推导、AlexNet Facial expression recognition system based on deep learning Abstract With the progress of society and the development of economy, artificial intelligence has been applied in a variety of scenarios, the most typical application is the application of robots. The design of human-computer interaction has become more and more mature. In order to understand the correct thoughts of people, robots should not only be reflected in language, but also analyze the correct emotions of people in other aspects. Facial expression recognition analysis is a major breakthrough point. This article from the nature of the neural network is introduced to use convolution neural network to build model for concrete analysis, including natural neural network and the convolution of the neural network knowledge, including some mathematical derivation, analysis on the structures, processes and results of the model, will be used in the knowledge points in detail summarized make people better understand and realize the actual model is set up. The model building in this paper is finally implemented by AlexNet, the champion of ILSVRC in 2012, and the model will be analyzed. At the end of the paper, it will also put forward the improvement of the model by using VGG network, ResNet, GoogLeNet and other methods with more layers of neural networks. Keywords: Facial expression recognition ; convolutional neural network; Knowledge of induction ; Mathematical deduction ; AlexNet ; 目 录 1 绪论 1.1 人脸表情识别的目的和现实意义 1 1.2 人脸表情识别技术国内外现状 1 1.3 环境的搭建 2 1.3.1 Pytho