| Since the 14 th Five-Year Plan,the country has focused on developing smart agriculture,establishing a big data system for agriculture and rural areas,and promoting the deep integration of new generation information technology with agricultural production and operation.Among them,the prevention and diagnosis of crop diseases is a very important aspect,and timely and effective diagnosis and treatment of diseases is an important means to stabilize the overall situation of agriculture and ensure stable and increased production of food crops in the current context of the continued spread of the global new crown pneumonia.For a long time,people have been judging whether crops suffer from diseases by empiricism and traditional image detection means,which is not only inefficient and the diseases are not timely and effectively prevented,but also consumes a lot of human and financial resources.In the context of the interaction between agriculture and information technology,more and more intelligent farming technology is sinking to the grassroots level,dedicated to the stable production of crops,and taking the lead in the agriculture 4.0 stage.Among them,machine learning technology,as an important means to achieve precision agriculture,is constantly applied to the field of online identification of crop diseases,becoming an important tool for modern farm management and an important tool for daily maintenance and management of agricultural practitioners.Based on the idea of "improvement and innovation" in industrial engineering and management,this study designs and implements the "Crop Doctor" online crop disease diagnosis and treatment system based on Python language and Django framework,relying on information management and information system.In this paper,we have done the following work on the design and development of "Crop Doctor" online crop disease diagnosis and treatment system.(1)Research on disease diagnosis based on transfer learning.In this part,we have obtained and compiled a large amount of sample data from open-source datasets through extensive data review and literature research,and pre-processed the crop disease images by threshold segmentation,rotation and translation,and scaling adjustment.Under the Pytorch deep learning framework,four pre-trained models,Res Net 18,Res Net 34,VGG 16 and Inception V3,were trained by migration learning based on the pre-processed dataset,and two crop disease diagnosis models were selected through training data and score screening.The system is applied to the construction of "Crop Doctor" system to realize online diagnosis of 81 types of crop diseases and pests including 13 types of crops,and through the condensation of various books and professional materials,the system of agricultural plant protection knowledge manual is constructed and applied to the plant protection information tips and plant protection information knowledge map function after disease diagnosis.(2)Development of "Crop Doctor" Web system based on Django.The overall design of the online crop disease diagnosis and treatment system and the implementation of each module function were carried out by applying software engineering methods and Django development framework.There are four main modules in the crop disease online diagnosis and treatment system: user login and registration module,crop disease online diagnosis module,plant protection information knowledge map module,and graphic display module,which are designed to realize the functions of one-stop disease detection and diagnosis,one-click access to plant protection information manual,and user registration and login.The integration of diagnosis and treatment is the responsibility and significance of the "Crop Doctor" system.The system aims to provide agricultural workers and agricultural researchers with a one-stop service platform for crop diseases and play a positive role in the research and development of precision agriculture and lightweight systems in the agricultural field. |
