北京理工大学珠海学院 2016 届本科生毕业设计基于深度学习的人脸表情识别系统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 北京理工大学珠海学院 2016 届本科生毕业设计Facial expression recognition system based on deep learningAbstractWith 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 北京理工大学珠海学院 2016 届本科生毕业设计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...