• 首页
  • 期刊简介
  • 编委会
  • 投稿指南
  • 收录情况
  • 杂志订阅
  • 联系我们
引用本文:李文逵,韩俊英.基于一种轻量级卷积神经网络的植物叶片图像识别研究[J].软件工程,2022,25(2):10-13.【点击复制】
【打印本页】   【下载PDF全文】   【查看/发表评论】  【下载PDF阅读器】  
←前一篇|后一篇→ 过刊浏览
分享到: 微信 更多
基于一种轻量级卷积神经网络的植物叶片图像识别研究
李文逵,韩俊英
(甘肃农业大学信息科学技术学院,甘肃 兰州 730070)
448671882@qq.com; 3243727977@qq.com
摘 要: 对轻量级卷积神经网络MobileNet V2的模型结构进行改进,将深度可分离卷积中的激活函数ReLU替换成Leaky ReLU,从而避免神经元死亡问题,倒置残差卷积中的跨越连接添加Dropout层,增大模型的泛化能力。实验结果表明,预测结果的总体准确率达到91.41%,最高精确率为95.12%,最高召回率为97.39%,取得较好的预测结果。这说明将MobileNet V2卷积神经网络用于植物叶片图像识别是实际可行的,为移动端植物叶片图像识别提供了实现方法和技术支撑。
关键词: 植物叶片;图像识别;MobileNet V2;卷积神经网络;深度学习
中图分类号: TP520.40    文献标识码: A
基金项目: 甘肃省自然科学基金资助项目(20JR5RA023).
Research on Plant Leaf Image Recognition based on a Lightweight Convolutional Neural Network
LI Wenkui, HAN Junying
(College of Information Science and Technology, Gansu Agricultural University, Lanzhou 730070, China )
448671882@qq.com; 3243727977@qq.com
Abstract: This paper proposes to improve the model structure of the lightweight convolutional neural network MobileNet V2. The activation function ReLU in the deep separable convolution is replaced with Leaky ReLU, thereby avoiding the problem of neuron death. A Dropout layer is added across connections in inverted residual convolution to increase the generalization of the model. The experimental results show that the overall accuracy rate of the prediction results reaches 91.41%, the highest accuracy rate is 95.12%, and the highest recall rate is 97.39%, achieving good prediction results. It shows that it is practical to use the MobileNet V2 convolutional neural network for plant leaf image recognition, and it provides an implementation method and technical support for mobile terminal realization of plant leaf image recognition.
Keywords: plant leaf; image recognition; MobileNet V2; convolutional neural network; deep learning


版权所有:软件工程杂志社
地址:辽宁省沈阳市浑南区新秀街2号 邮政编码:110179
电话:0411-84767887 传真:0411-84835089 Email:semagazine@neusoft.edu.cn
备案号:辽ICP备17007376号-1
技术支持:北京勤云科技发展有限公司

用微信扫一扫

用微信扫一扫