摘要
棉花作为我国最主要的农产品之一,不仅具有不错的观赏价值,更重要的还是工业原料。棉花的花型不同于其他花卉种类,且不同种类其纤维长度还有所差异。为了解决棉花人工区分效率低的问题,本文基于深度学习方法,以棉花原始的图像数据作为研究对象,通过多层网络学习棉花的特征信息,更加精确区分不同类型的棉花种类。试验结果表明:本文所提出的卷积神经网络CNN-CSC模型相较于传统机器学习方法识别精度提升大约15%,平均精度达到89.17%,为棉花的自动化管理提供了一种有效的手段。
Cotton,as one of the most major agricultural products in our country,not only has a good ornamental value,but also is more important and an industrial raw material.The flower type of cotton is different from those of other floral species,and the fiber lengths of the different species also differ.To solve the problem of low efficiency of artificial discrimination of cotton,in this paper,based on deep learning method,the original image data of cotton is used as a research object to learn the characteristic information of cotton through multi-layer network and distinguish different types of cotton species more precisely.Test results:the cnn-csc model of the convolutional neural network proposed in this paper improves the identification accuracy by approximately 15%and the average accuracy reaches 89.17%compared to traditional machine learning methods,which provides an effective means for automated management of cotton.
作者
李海涛
罗维平
LI Hai-tao;LUO Wei-ping(School of Mechanical Engineering and Automation,Wuhan Textile University,Wuhan Hubei 430200,China)
出处
《武汉纺织大学学报》
2022年第4期22-26,共5页
Journal of Wuhan Textile University
关键词
深度学习
卷积神经网络
图像识别
棉花识别
deep learning
convolution neural network
image recognition
cotton recognition