摘要
卷积神经网络出现之前图像识别方法主要依赖人工设计特征,而这样的特征只能表征图像中的中低级信息,难以提取图像的深层次信息.卷积神经网络通过建立深度神经网络来模拟人脑分析、学习和解释数据,具有强大的表达能力和泛化能力,能够更好地表示图像的深层次信息.开展基于卷积神经网络对图像识别进行研究可以推动计算机领域的发展.本文先对卷积神经网络做一个概述,重点综述了卷积神经网络相关算法在人脸识别、人体动作识别、医疗图像处理和农业病虫害识别方面的应用及其优缺点,最后探讨了卷积神经网络在图像识别上所面临的挑战和展望.
Before the advent of convolutional neural networks image recognition methods mainly relied on artificially designed features,and such features can only represent the low-level information in the image,and it is difficult to extract the deep-level informa—tion of the image.A convolutions!neural network simulates the human brain to analyze,learn and interpret data by establishing a deep neural network.It has strong expression and generalization capabilities and can better represent the deep-level information of the image.Research on image recognition based on convolutional neural networks can promote the development of the computer field.This article first gives an overview of convolutional neural networks,focusing on the applications and advantages and disadvantages of con—volutional neural network-related algorithms in face recognition,human action recognition,medical image processing,and agricultural pests and diseases recognition,and finally discusses convolutional neural networks.Challenges and prospects faced by the network in image recognition.
作者
盖荣丽
蔡建荣
王诗宇
仓艳
陈娜
GAI Rong-li;CAI Jian-rong;WANG Shi-yu;CANG Yan;CHEN Na(College of Information Engineering,Dalian University,Dalian 116622,China;University of Chinese Academy of Science,Beijing 100049,China;Shenyang Institute of Computing Technology,Chinese Academy of Sciences,Shenyang 110168,China)
出处
《小型微型计算机系统》
CSCD
北大核心
2021年第9期1980-1984,共5页
Journal of Chinese Computer Systems
基金
大连市科技创新基金项目(2020JJ27SN101)资助。
关键词
卷积神经网络
图像识别
人体动作识别
人脸识别
convolutional neural network
image recognition
human action recognition
face recognition
