%0 Journal Article
%T 基于卷积神经网络的国兰种类识别系统<br>Chinese Orchid Species Recognition System Based on Convolutional Neural Network
%A 张子赫
%A 赵吟吟
%J Computer Science and Application
%P 2346-2353
%@ 2161-881X
%D 2020
%I Hans Publishing
%R 10.12677/CSA.2020.1012248
%X <div style=\"text-align:justify;\">
	<span style=\"font-size:14px;\">针对兰花品种众多，外表相似，导致没法准确识别兰花的种类这一问题，本文构建了基于卷积神经网络的国兰识别模型，设计并实现了模型在移动端的应用。文中通过多种途径完成常见国兰数据集的创建，进而以Inception-ResNet-v2为卷积神经网络预训练模型，使用迁移学习技术完成模型训练，并基于Android平台完成系统的开发和测试。测试结果显示，对传统国兰图像分类识别准确率达到91.51%。<br />
Aiming at the problem that there are many varieties of orchids with similar appearances, which makes it impossible to accurately identify the types of orchids, a Chinese orchid recognition model based on convolutional neural network is constructed, and its transplantation to the mobile terminal is designed and implemented. First, this paper completes the creation of the common Chinese orchid data set through multiple channels, and then uses the Inception-ResNet-v2 convolutional neural network as the pretraining model, uses the transfer learning technology to complete the model construction, and develops and tests the system based on the Android platform. Through experiments and actual tests, the accuracy of classification and recognition of traditional Chinese orchid images reached 91.51%.</span> 
</div>
%K 国兰，Inception-ResNet-v2网络，图像识别，Android<br>Chinese Orchid
%K Inception-Resnet-V2 Network
%K Image Identification
%K Android
%U http://www.hanspub.org/journal/PaperInformation.aspx?PaperID=39405