Research And Development Of Tobacco Pest Identification And Management System Based On Convolutional Neural Network

Shandong is the main tobacco producing area in China,with good quality of tobacco leaves,which has created a stable tax for the country and a stable income for tobacco farmers.However,the problem of insect pest has been perplexing the production activities of tobacco farmers.The development of Internet of things and artificial intelligence technology provides convenience for scientific remote accurate and efficient identification,diagnosis and control of tobacco pests,which is of great significance and role in ensuring tobacco health and tobacco yield.In this paper,image recognition technology was used to study the main tobacco pests,and tobacco pest management system was developed,which reduced the labor intensity of tobacco farmers’ pest management and control:(1)Construction of tobacco pest identification modelThe image data sets of tobacco pests were acquired by field collection and web crawler crawling.The images of pests were collected by manual photographing;Web crawler can get the image of tobacco pests by crawling the data on the Internet.Four kinds of pests,including cotton bollworm adults,cotton bollworm larvae,Spodoptera litura and Myzus persicae,were selected as the data sets of this experiment.The pest data set was preprocessed by image flipping,image graying and histogram equalization.The image size was adjusted to 50×50 pixels by image size normalization;Based on Alex Net,Google Net and VGG-19 network,a tobacco pest recognition model was constructed.The input image was extracted with multi-dimensional and multi-scale features through convolution layer,and the main features were extracted through pooling layer;Finally,the features are input into the classifier to recognize and classify the tobacco pest images,and the pest types are obtained to realize the automatic recognition of tobacco pest images.(2)Internet of things monitoring system for tobacco field environmental informationAnalyze the environmental factors in the process of tobacco growth,among which air temperature and light intensity have the highest impact on pests.In order to monitor the temperature and light intensity information of tobacco field in real time,the system designs the Internet of things intelligent environmental monitoring system,which includes the Internet of things intelligent tobacco environmental information sensing module,transmission module and host computer monitoring system module,Finally,the function of real-time monitoring tobacco growth environment information is realized.(3)Tobacco pest management systemCombined with the user’s functional requirements,the tobacco pest management system was designed and developed.My SQL database was selected to design the system database.Bootstrap framework and SSH framework were used to develop the system.The functions of tobacco pest information query,pest diagnosis,pest identification,pesticide information query,expert online consultation and real-time monitoring were realized.In this study,the image recognition model of tobacco pests was constructed,and the tobacco pest management system was developed.The functions of tobacco pest query,pest diagnosis,pest recognition,pesticide information query,expert online consultation and realtime monitoring were realized.The wechat program of tobacco pest image recognition is developed,and the function of online recognition of tobacco pest image is realized.It can reduce the cost of artificial field diagnosis,enhance the timeliness and real-time of tobacco pest control,and improve the income of tobacco farmers;It provides important theoretical and technical support for accurate and efficient identification,diagnosis and control of tobacco pests.