Recognition And Classification Of Fresh Tea Leaves Based On Deep Learning

China is the largest tea producer in the world.In the current tea production,fresh tea leaves processed by traditional industrial tea fresh leaf classifiers have problems such as poor classification effect and inconvenient collection.This is not conducive to improving the economic benefits of tea.This study is aimed at identifying and classifying fresh tea leaves based on machine-extracted tea leaves,combined with deep learning,through specific experiments.The main work of this article is as follows:(1)The research background and significance of this article are introduced.The research status of domestic and foreign research on the classification and deep learning of fresh tea leaves and related theoretical basis are studied.The layers of the convolutional neural network and the related key technical principles are studied.(2)A small sample data set of fresh tea leaves is established.Area cropping,graying,size normalization and other methods are used to highlight the characteristic information of fresh tea leaf images.The original tea fresh leaf small sample data set and the tea fresh leaf picture data set enhanced by image data are completed.(3)The VGG16,Inception-v3,Res Net-50 and Mobile Net-v2 models in convolutional neural networks are studied in detail.Experiments were performed on the fresh tea leaf data set using the above network model and related parameters were optimized for comparison.After comparison,the VGG16 model that is most suitable for the fresh tea leaf dataset is selected,and then the network is trained using transfer learning and data augmentation techniques.Good results have been achieved through experiments.