基于MobileNet V2迁移学习的中药材图像识别-豆柴文库

您所在位置：网站首页 / 基于MobileNet V2迁移学习的中药材图像识别.docx / 文档详情

在线预览结束，喜欢就下载吧，查找使用更方便

5 金币

下载文档

/ 2

下载提示

如果您无法下载资料，请参考说明：

1、部分资料下载需要金币，请确保您的账户上有足够的金币

2、已购买过的文档，再次下载不重复扣费

3、资料包下载后请先用软件解压，在使用对应软件打开

文本预览

基于MobileNetV2迁移学习的中药材图像识别
Title:TransferLearning-BasedChineseHerbalMedicineImageRecognitionusingMobileNetV2
Introduction:
ChineseherbalmedicinehasbeenusedforcenturiesintraditionalChinesemedicineforvariouspurposes,suchaspreventingandtreatingdiseases,promotingoverallhealth,andimprovingthebody'snaturaldefensemechanisms.Withtheadvancementoftechnology,computervisionandmachinelearningtechniqueshaveemergedaspowerfultoolsforautomatingtheidentificationofChineseherbalmedicines.Inthispaper,weproposeatransferlearning-basedapproachusingMobileNetV2forChineseherbalmedicineimagerecognition.
1.LiteratureReview:
Severalstudieshavebeenconductedonplantrecognitionandidentificationusingcomputervisiontechniques.However,veryfewstudiesfocusspecificallyonChineseherbalmedicinerecognition.Transferlearning,atechniquethatutilizestheknowledgelearnedfromonedomaintoimprovelearninginanotherdomain,hasshownpromisingresultsinimagerecognitiontasks.MobileNetV2,alightweightconvolutionalneuralnetwork(CNN)architecture,hasgainedattentionforitsefficiencyandhighaccuracy.ThispapercombinestransferlearningwithMobileNetV2tobuildarobustChineseherbalmedicinerecognitionmodel.
2.DataCollectionandPreprocessing:
AlargedatasetofChineseherbalmedicineimagesiscollected,consistingofhigh-qualityimagesofdifferentherbalmedicines.Theseimagesareobtainedfromreliablesourcesandarelabeledwiththecorrespondingherbnames.Toensuretheaccuracyofthedataset,expertsinChineseherbalmedicinevalidatethelabels.Theimagesarepreprocessedbyresizingthemtoasuitablesize,normalizingpixelvalues,andaugmentingthedatatoincreasethediversityofthedataset.
3.TransferLearningApproach:
Transferlearningisappliedusingthepre-trainedMobileNetV2model,whichistrainedonthelarge-scaleImageNetdataset.ThelastoutputlayerofMobileNetV2isreplacedwithanewfullyconnectedlayerwithsoftmaxactivationforherbalmedicineclassification.Onlytheweightsofthenewlyaddedlayeraretrained,whiletheweightsoftherestofthenetworkarefrozen.Thisapproachallowsthemodeltobenefitfromthepre-existingknowledgeofMob