摘要
我国作为人口大国,对粮食的需求量与日俱增。文章以水稻田间管理为研究对象,采集广东省内不同地区的水稻田图像,基于深度学习技术,采用多分类器融合的思想,将不同分类器联合使用,得出以下结论:直接应用传统CNN网络辨识准确率低,采用多分类器融合的思想,得到的分类结果准确性高,融合后的分类性能优于任何单一分类器的分类性能,所提出的方法11类总体分类准确率达到70.3%。
As a country with a large population,China's demand for food is increasing day by day.This paper takes rice field management as the research object,collects paddy field images from different regions in Guangdong Province,uses the idea of multi-classifier fusion based on the deep learning technology,and uses different classifiers together,and draws the following conclusions:the accuracy of identification by directly applying traditional CNN network is low,and the accuracy of classification results by using the idea of multi-classifier fusion is high,and the classification performance after fusion is better than that of any single classifier,The overall classification accuracy of 11 categories of the proposed method is 70.3%.
作者
张勤学
潘明海
陈俊聪
汪昊
张誉铧
ZHANG Qinxue;PAN Minghai;CHEN Juncong;WANG Hao;ZHANG Yuhua(Zhonghe Technology(Guangdong)Co.,Ltd.,Guangzhou 510000,China)
出处
《数字通信世界》
2023年第7期49-51,共3页
Digital Communication World