摘要
通过采集到的农作物叶片图像对病害进行识别具有重要的意义。为能够实现通过采集到的农作物叶片图像对病害进行快速识别,笔者提出一种基于Mobilenet的农作物叶片病害识别方法,利用轻量化卷积神经网络Mobilenet作为特征提取器。该方法在一般CPU上的推理时间为5 ms,在Plant Village数据集上的测试精度达到99.57%。
It is of great application significance to identify diseases through the crop leaf images. In order to realize rapid and accurate disease recognition through the collected crop leaf images, a crop leaf disease recognition method based on mobilenet is proposed. Using the lightweight convolution neural network mobilenet as the feature extractor, combined with the migration learning strategy, the disease can be quickly recognized through the crop leaf image. The reasoning time of this method on general CPU is 5 ms,and the test accuracy on plant Village data set is 99.57%.
作者
秦嘉奇
QIN Jiaqi(Guilin Institute of Information Technology,Guilin Guangxi 541000,China)
出处
《信息与电脑》
2021年第18期181-184,共4页
Information & Computer
基金
2020年度广西高校中青年教师科研基础能力提升项目(项目编号:2020KY57020)。
关键词
病害检测
轻量化CNN
细粒度分类
深度学习
disease detection
lightweight CNN
fine grained classification
deep learning