Diagnosis And Control System Of Tomato Diseases And Pests Based On Deep Learning

There are 24 provinces in China that grow tomatoes on a large scale.In the past 10years(2010-2019),the production of tomato has been growing positively in China,with an annual scale of over 60 million tons.However,tomato pests and diseases have become a major factor restricting the development of tomato industry.In the process of tomato growth,due to pests,viruses and other factors,its yield is reduced;due to improper control measures,its quality is reduced.The traditional manual diagnosis method has poor reliability and wastes human resources;the prevention and control measures are mainly the use of pesticides,which have low safety and poor pertinence.With the development of smart agriculture,diagnosis methods based on deep learning are replacing traditional manual diagnosis methods.Therefore,some diagnosis systems based on deep learning have been developed.However,some existing diagnosis systems have large model parameters and slow convergence speed.The diagnosis accuracy is not ideal,the generalization ability is weak,and there is no relevant prevention knowledge base.In order to solve the above-mentioned problems in the current diagnosis and control of tomato diseases and pests based on deep learning,this paper designed a lightweight convolutional neural network with multi-size convolution kernel and maximum pooling,which reduced the amount of parameters to a certain extent and improved the accuracy of diagnosis.Secondly,in order to further improve the diagnostic accuracy,Res Net50 is selected for migration learning,and the final output of the convolutional layer is batch normalized,and the global average pooling operation was performed before the custom fully connected layer,which improved the convergence speed.It reduced the computational burden and enhances the generalization ability.Finally,the diagnosis model and the prevention and control knowledge base were combined to develop a simple operation system through the Flask framework,and the tomato pest diagnosis and control system based on deep learning is realized,so that it can be applied to tomato Planting process.This paper designed and implemented a tomato disease and insect pest diagnosis and control system that can quickly and effectively diagnose tomato diseases and insect pests,give prevention suggestions,and popularize related prevention knowledgethrough the transfer learning of the convolutional neural network,which had solved the problem of tomato producers.In the process of planting,problems such as low accuracy of diagnosis and control of diseases and insect pests,incomplete knowledge of prevention and control,and improper prevention and control methods are encountered.The subsequent research on the diagnosis and control of tomato diseases and insect pests according to the degree of infection has certain reference significance and promotes the development of smart agriculture.