Intelligent Classification of Tobacco Leaves based on Residual Network

Aiming at the problem of intelligent grading of tobacco leaves in the automatic tobacco grading system, a method of intelligent grading of tobacco leaves based on deep residual network is designed. The method includes: reading the tobacco leaf image data set, preprocessing the images in batches, inputting the processed image data into the grading training model, and finally outputting the grading result of the tobacco leaf after the model prediction. This paper builds a custom training model based on the convolutional residual neural network to intelligently classify tobacco leaves. The image enhancement function is added to the preprocessing module to perform preprocessing such as random rotation, shearing, scaling, etc., to expand the polymorphism of tobacco leaf samples and improve the compatibility of the model. Regularization and Dropout layers are added to the convolutional neural network module to solve the phenomenon of over-fitting in the network.