中国农业科技导报2025，Vol.27Issue(4)：99-109,11.DOI:10.13304/j.nykjdb.2023.0785
基于卷积神经网络的农作物病害识别研究
Research Progress on Crop Diseases Identification Based on Convolutional Neural Network
摘要
Abstract
Crop diseases are major threats for agricultural production,so timely and accurate identification of disease is important for the development of control measures to ensure food security.With the rapid development of deep learning,convolutional neural networks are used more and more to identify crop diseases.This paper compared the advantages and disadvantages of convolutional neural network disease recognition methods from 3 aspects including disease recognition based on different data sets,disease recognition using transfer learning and pre-training,and lightweight of the disease recognition model.It also analyzed the shortcomings of the current methods and put forward the future development trend.It was pointed out that more abundant data sets should be constructed,multi-modal data should be combined,models should be further optimized,and robots should be used to implement automatic detection.It provided important references for reducing food loss,realizing precision agriculture management,promoting agricultural modernization and sustainable development.关键词
深度学习/卷积神经网络/农作物病害/识别Key words
deep learning/convolutional neural network/cropdiseases/identification分类
农业科技引用本文复制引用
陈自立,林卫,贺佳,王来刚,郑国清,彭一龙,焦家东,郭燕..基于卷积神经网络的农作物病害识别研究[J].中国农业科技导报,2025,27(4):99-109,11.基金项目
国家自然科学基金项目(41601213) （41601213）
国家重点研发计划项目(2022YFD2001105) （2022YFD2001105）
河南省重点研发与推广专项(232102111030,232102110027) （232102111030,232102110027）
河南省农业科学院自主创新项目(2023ZC064). （2023ZC064）
