Research on Rose Classification Based on Neural Network Model

Abstract

As an important industry in China, rose has extremely important economic value and ornamental value. However, affected by human factors and planting environment, even roses of the same variety will have different levels of classification. In order to solve the low efficiency of wrong classification and missing classification caused by manual visual classification and realize the transformation of rose classification from manual judgment to machine independent recognition, this paper first performs data enhancement, image naming and other operations on 200 manually photographed yellow rose images of 5 levels of the same variety, and constructs an image database. Then, in Python language and tensorflow2.0 deep learning tool, design rose classification algorithm based on artificial neural network(ANN), rose classification algorithm based on convolution neural network(CNN) and rose classification algorithm based on feature extraction(FE) and artificial neural network(ANN). Finally, the experimental results show that the third algorithm significantly speeds up the running speed and significantly improves the average recognition rate compared with the first two algorithms, which effectively solves the problems of wrong or missing classification and low classification efficiency in manual intuitive classification, and realizes the transformation from manual classification to machine independent classification.