基于机器学习与深度学习的地基云识别进展

打开文本图片集

中图分类号：TP391.41 文献标志码：A 文章编号：2095-2945（2025）18-0022-07

Abstract：Withtherapid developmentofdeeplearning technology，transfer learning hasalsobeensucesfullyappliedin thefieldofimageprocessingbasedondeeplearning.Inordertoverifythefeasibilityofdeeplearningtechnologycombnedwith transferleaninginground-basedcloudspeciesidentification，andtheefectivenessandadvancementcomparedwithtraditional machine learning methods，thispaperconductsmachine learningandtransferdeep learningonthe SWIMCATground-based clouddataset.Experimentalcomparisonoflearningmethods，experimentalcomparisonoftransferlearningandnon-transfer learning，andvisualanalysisofthecharacteristicsofground-basedcloudsinconvolutionaleuralnetworks.Theadvantagesnd disadvantagesofmachinelearningandtransferdeeplearninginground-basedcloudrecognitionarecomparedandanalyzed throughxperiments，layingatheoreticalfoundationforsubsequentin-depthresearchondeeplearningground-basedcloud recognition algorithms.

Keywords： deep learning; transfer learning; machine learning; ground-based cloud map; visualization

近年来，云及其特性的研究成为各学者的研究重点，尤其是在云检测与分类、云量估计、云移动轨迹预测以及云类型演变趋势预测等动态行为预测领域。（剩余10787字）