Research And Application Of Tobacco Maturity Determination Based On ResNet

Maturity is the primary indicator of tobacco leaf quality.At present,the tobacco leaf maturity is mainly identified by manual,which has problems such as low efficiency and strong subjectivity.Therefore,an efficient and accurate new method is urgently needed to replace the traditional manual identification method.Nowadays,with the rapid development of deep learning,Res Net has been widely used in many fields because of its advantages of improving the network depth and solving the problem of network gradient disappearance.Therefore,it is of great significance to apply Res Net to tobacco leaf maturity determination.The existing research on tobacco leaf maturity determination has problems such as low accuracy and a single experimental dataset,which cannot be applied to the actual scene.This paper studies the tobacco leaf maturity determination model based on Res Net,and designs and implements a tobacco leaf maturity analysis system.Firstly,this paper designs and implements a tobacco leaf image acquisition system to obtain the dataset required for the experiment.Compared with the existing tobacco leaf dataset,the dataset in this paper is larger in scale and richer in varieties and parts.In the process of dataset preprocessing,a 24-color standard color card is introduced for image color correction,which can improve the image quality of the dataset and has good generality.Then,the tobacco leaf maturity determination model was established based on Res Net18,Res Net34,and Res Net50,and the model was trained under the optimizer SGDM,RMSProp,and Adam respectively.Compared with the existing best method VGG16 Net,the experimental results show that the accuracy of the three models proposed in this paper is higher than that of VGG16 Net,indicating that the method in this paper can effectively improve the accuracy of tobacco leaf maturity determination.Finally,in order to apply this method to the actual scene,a tobacco leaf maturity analysis system is designed and implemented,which integrates the preprocessing algorithm of the tobacco leaf image and the tobacco leaf maturity determination model.The test shows that the system can realize the real-time analysis of tobacco leaf maturity,which has important practical significance.