A fruit image classification method and device based on a neural network and transfer learning

The invention relates to a fruit image classification method and device based on a neural network and transfer learning. The method comprises the following steps: preprocessing the image to be classified, preprocessing the image, and then combining the SMOTE algorithm for data enhancement; Carrying out BN batch normalization operation on pixel points of the data enhanced image to make the pixel points conform to a normal distribution; The pixel points of the normal distribution are inputted into a fruit image classification model based on neural network and transfer learning, and the classification result of the image is outputted. This method can be used to classify the input images of arbitrary size. The method relies on the classification model of fruit image based on neural network andtransfer learning, which improves the efficiency of recognition and classification, reduces a lot of time cost and has high reliability. 本发明涉及种基于神经网络和迁移学习的水果图像分类方法及装置，该方法包括：将待分类的图像进行预处理，经预处理后再结合SMOTE算法进行数据增强；将数据增强后的图像的像素点进行B