摘要
深度学习(Deep Learning,DL)已广泛应用于智能农业的病虫害检测、植物和水果识别、农作物及杂草检测与分类等研究中。对2014年至2019年国内发表的65篇有关DL在农业中应用研究成果进行综述。简要介绍DL的基本概念及其发展历史,给出了所选论文检索方法及其分布;对所选论文从研究对象与目的、数据来源、类间差异、预处理、数据扩增、模型框架以及性能对比等角度进行了综述;对DL的优缺点进行了分析,并指明了其在智能农业研究中的发展趋势。
Deep Learning(DL)has been widely used in intelligent agriculture for plant disease detection, plant and fruit recognition, crop and weed detection and classification and so on. 65 articles from 2014 to 2019 on the application of DL in agriculture of China are presented. At first, the basic concept and development history of DL are briefly introduced, and the article retrieval and distribution of the reviewed articles are given. Subsequently, the articles are reviewed from various points of view such as research object, data source, inter-class differences, preprocessing, data-augmentation, framework and performance comparison. Eventually, the advantages and disadvantages of DL are analyzed, and its development trends in agriculture are demonstrated.
作者
吕盛坪
李灯辉
冼荣亨
LV Shengping;LI Denghui;XIAN Rongheng(College of Engineering, South China Agricultural University, Guangzhou 510642, China)
出处
《计算机工程与应用》
CSCD
北大核心
2019年第20期24-33,51,共11页
Computer Engineering and Applications
基金
国家自然科学基金（No.51605169）
关键词
深度学习
智能农业
检测
识别
分类
预测
deep learning
intelligent agriculture
detection
recognition
classification
prediction
作者简介
通讯作者:吕盛坪(1982-),男,博士,副教授,主要从事农业/工业数据挖掘、机器学习、深度学习研究,E-mail:lvshengping@scau.edu.cn。
