Research And Implementation Of Crop Pest Identification System Based On Improved Convolutional Neural Network

In recent years,with the constant changes in the international situation,countries all over the world pay more attention to food security.Rice is the first of the three main grains and the cornerstone of food security.How to promptly prevent and defuse the risks of crop diseases and insect pests is an important link in crop security work.In the prevention of diseases and insect pests,the most critical step is to find and accurately identify and timely treat diseases and insect pests.The traditional pest and disease identification system is very complicated for the image preprocessing process,the selection of image features is not accurate and the recognition accuracy is not high,which to a certain extent interferes with the agricultural pest experts to timely control the work of pest and disease.How to improve the identification accuracy of pest and disease identification system also has high practical application value for preventing and resolving crop disease and pest risks.Based on the analysis of relevant research status at home and abroad,and based on improved AlexNet and Inception models,an improved convolutional neural network disease and insect identification model is proposed and designed.The model trains Inception module by transferring the image feature knowledge learned by AlexNet.The advantages of the two networks are fully combined to improve the accuracy of recognition.In addition,the problem of insufficient samples of crop data sets is effectively solved through data set enlargement of image data sets,so that the training data sets are effectively enlarged.The results are summarized through comparative experiment analysis.The improved convolutional neural network model in this paper has a better effect on the accuracy of pest and disease identification,which verifies the superiority of the model.Based on the above work,in this paper,an improved convolutional neural network based crop pest identification system is designed and implemented.Based on actual needs,this system can be oriented to agricultural disease and pest experts and ordinary users,and has functions such as login registration,data management,system management,image recognition,expert verification and analysis.After users upload pictures of diseases and pests,the system extracts the features of the pictures.The identification results and relevant solutions will be returned in the end.At the same time,agricultural disease and pest experts will be supported to test the identification results and modify the identification results.The system is developed based on python,the front page uses layui,the database used by the system is MySQL database,and the data set uses the picture samples of crop diseases and pests in Plant Village.Through the functional test and non-functional test of the system,the system can efficiently complete the identification of crop diseases and pests,and meet the design standards in function and performance.