Research and Implementation Of Image Recognition of Tea Based on Deep Learning

Abstract

With the advent of the Internet era, deep learning technology is developing rapidly in a wide range of application scenarios. As one of the main research methods of deep learning, convolutional neural network is widely used in the field of image recognition [5]. This paper hopes to build a modern intelligent tea image recognition system by combining in-depth learning technology, so as to help inherit traditional Chinese culture. The main work of this paper is as follows:(1)Making a sample data set containing about 49000 pictures;(2)Constructing convolution neural network;(3)The correct rate of training image recognition model is 96%;(4)Realize tea picture recognition;(5)Build the web page.Under the framework of tensorflow, the program of this subject realizes the tea picture recognition function by training convolution neural network [5][6], and builds it in the beautiful web page, so that the tea culture can be presented and spread in a more modern form, which is conducive to the inheritance and development of Chinese traditional culture.