摘 要：在图像分类、目标检测等领域的应用前景非常可观。然而，卷积神经网络依然存在着过拟合、梯度消失等问题。鉴于此，文章首先介绍了卷积神经网络的发展历程以及经典的网络模型。其次具体分析了各种卷积神经网络的结构和优缺点，并针对以上问题给出了相应的解决方法。最后分析了卷积神经网络在图像分类领域的不足并展望了未来的发展方向。
关键词：深度学习；卷积神经网络；图像分类；计算机视觉；过拟合
DOI:10.19850/j.cnki.2096-4706.2022.16.008
中图分类号：TP181 文献标识码：A 文章编号：2096-4706（2022）16-0029-04
A Review of the Application of Deep Learning in Image Classification
JIN Wei 1, MENG Xiaoman2, WU Yichao3
(North China University of Water Resources and Electric Power, Zhengzhou 450046, China)
Abstract: In recent years, deep learning has been widely used in the field of computer vision, and convolutional neural network is also one of the more important research directions in this field. Convolutional neural network has a promising application in image classification, object detection and other fields. However, convolutional neural networks still have problems such as over fitting and gradient disappearance. In view of this, this paper first introduces the development of convolutional neural network and the classical network model. Secondly, the structure, advantages and disadvantages of various convolutional neural networks are analyzed in detail, and the corresponding solutions to the above problems are given. Finally, the shortcomings of convolutional neural network in the field of image classification are analyzed and the future development direction is prospected.
Keywords: deep leaning; convolutional neural network; image classification; computer vision; over fitting
参考文献：
[1] GUO Y M，LIU Y，OERLEMANS A，et al. Deep Learning for Visual Understanding: A Review [J].Neurocomputing，2016，187：27-48.
[2] LECUN Y，BOTTOU L，BENGIO Y，et al. Gradient-Based Learning Applied to Document Recognition [J]. Proceedings of IEEE， 1998，86（11）：2278-2324.
[3] CHEN R，WANG M L，LAI Y. Analysis of the Role and Robustness of Artificial Intelligence In Commodity Image Recognition Under Deep Learning Neural Network [J]. Plos ONR，2020，15（7）：e0235783.
[4] 田萱，王亮，丁琪 . 基于深度学习的图像语义分割方法综述 [J]. 软件学报，2019，30（2）：440-468.
[5] ZHOU L N，PAN S M，WANG J W，et al. Machine Learning on Big Data: Opportunities and Challenges [J].Neurocomputing， 2017，237：350-361.
[6] SHAO L，WU D，LI X L. Learning Deep and wide：a Spectral Method for Learning Deep Detworks [J].IEEE Transactions on Neural Networks and Learning Systems，2014，25（12）：2303-2308.
[7] BARLOW H B. Unsupervised learning [J].Neural computation，1989，1（3）：295-311.
[8] HINTON G E. A Practical Guide to Training Restricted Boltzmann Machines [J].Momentum，2012，9（1）：599-619.
[9] HAWKINS D M. The Problem of Overfitting [J].Journal of Chemical Information and Computer sciences，2004，44（1）：1-12.
[10] KRIZHEVSKY A，SUTSKEVER I，Hinton G E. Imagenet Classification with Deep Convolutional Neural Networks [J]. COMMUNICATIONS OF THE ACM，2017，60（6）：84-90.
[11] HE K M，ZHANG X Y，REN S Q，et al. Deep Residual Learning for image Recognition [J].IEICE Transactions on Fundamentals of Electronics, Communications and Computer Sciences，2016：770-778.
[12] FRAZIER-LOGUE N, HANSON S J. Dropout is a Special case of the Stochastic Delta Rule: Faster and more accurate Deep Learning [J/OL].arXiv:1808.03578v2 [cs.LG].[2022-05-06].https://arxiv. org/pdf/1808.03578.pdf.
[13] SIMONYAN K, ZISSERMAN A. Very Deep Convolutional Networks for Large-scale image Recognition [J/OL].arXiv:1409.1556v6 [cs.CV].[2022-05-02].https://arxiv.org/pdf/1409.1556.pdf%E3%80%82.
[14] SZEGEDY C，LIU W，JIA Y Q，et al. Going Deeper with Convolutions[EB/OL].[2022-05-04].https://www.cv-foundation. org/openaccess/content_cvpr_2015/papers/Szegedy_Going_Deeper_ With_2015_CVPR_paper.pdf.
[15] HE K M，ZHANG X Y，REN S Q，et al. Deep Residual Learning for image Recognition [EB/OL].[2022-05-01].https:// openaccess.thecvf.com/content_cvpr_2016/papers/He_Deep_Residual_ Learning_CVPR_2016_paper.pdf.
[16] HUANG G，LIU Z，MAATEN L V D，et al. Densely Connected Convolutional Networks [EB/OL].[2022-04-29].https:// openaccess.thecvf.com/content_cvpr_2017/papers/Huang_Densely_ Connected_Convolutional_CVPR_2017_paper.pdf.
作者简介：金玮（1996—），男，汉族，河南周口人，硕士研究生在读，研究方向：图像分类识别；孟晓曼 (1998—)，女，汉族，河南洛阳人，硕士研究生在读，研究方向：点云语义分割；武益超（1999—），男，汉族，河南安阳人，硕士研究生在读，研究方向：点云语义分割。