基于卷积神经网络的花卉图像分类研究.pdf

基于卷积神经网络的花卉图像分类研究 161203107240 · 2020 的基于卷积神经网络的花卉图像 分类研究的研究的 的 的 2020 5 1 2020 基于卷积神经网络的花卉图像分类研究 花卉分类图像分类的研究花 的类的分类的研究 的的Oxford 17 Flowers 花卉Python Tensorflow 花卉图像分类经神经网络的基卷 积神经网络CNN , 图像, 图像分类的的 68%0-16 的花卉分类17 类花 图像分类 Tensorflow 卷积神经网络 2020 Study onflower image classification basedonconvolutionalneural network Abstract Flowerclassificationhasalwaysbeenoneoftheimportantresearchpointsofimageclassi- ficationindeeplearning,whichismainlycausedbythesimilarityofflowersandthediffer- ence between classes, and has always been the research direction of scientists. This paper selectsthefamousOxford17FlowersdatasetandUSEStheTensorflowmoduleinPython software to realize the flower image classification problem. Firstly, this paper constructs the CNN model of convolutional neural network on the basis of classical neural network to extract the features of the image. Then, the designed model is used to extract the target featuresfromcolor,brightness,shapeandotherfeatures,andthegradientdescentalgorithm is used for image recognition and classification. Finally, the verification accuracy reached 68%,andthedatawereaccuratelyclassifiedaccordingtotheChineseflowernamelabelof 0-16, withatotalof17types offlowers. Keywords:Ima