Research And Implementation Of Pest Detection Algorithm And Mobile Client Based On Convolutional Neural Network

With the development of social science and technology and economy in today's society,traditional agricultural planting and management models can no longer meet the needs of today's agricultural development.The traditional agricultural model must be reformed by informationization.Therefore,the design and development of an agricultural remote monitoring system integrating intelligence,information,and convenience is particularly important for the development of modern agriculture.There are many agricultural remote monitoring systems that have been put into operation in the market,mostly based on the PC side.The number of mobile clients is limited and their functions are limited.Most of them are based on temperature and humidity detection and lack of pest and disease detection functions.Therefore,this paper analyzes the design and implementation of the entire agricultural remote monitoring system and carries out the following two parts of the study:(1)Focus on the design and implementation of the system client so that users can grasp the growing information of crops anytime,anywhere,and perform effective and accurate remote supervision on crops.This system fully integrates the mobility and convenience of Android system.The design and implementation of the agricultural remote monitoring system mobile client includes core function modules such as environment detection module,video monitoring module,equipment control module,and pest and disease detection module.The design of the mobile client adopts the modular design and uses the MVC design mode to separate the data and view levels of each module in the program,and to improve the program scalability and maintainability.In the design of the interface,the use of Fragment to build a tab-tabbed sliding orientation layout interface is easy to use and highly applicable.The SQL Server database corresponding to the agricultural remote monitoring system is designed to store various key information.The server uses multi-threading technology to take full advantage of the advantages of multi-core CPUs,making the server highly capable of concurrent processing.(2)In order to further improve the function of the entire system,the pest and disease recognition algorithm based on convolutional neural network was deeply studied and applied to the pest and disease detection module.Compared with the traditional image processing technology and basic classifiers,this algorithm has great advantages.Firstly,it uses the Haar feature extraction algorithm in image processing and the classical Adaboost algorithm in machine learning to determine whether crops contain crop fruits.Judgment,this can filter the missing video information and extract the samples needed for the experiment,so as to obtain pictures containing crops,to facilitate the collection of crop samples;the second is to test the accuracy of the test set to achieve 98.8%while training classification The convolutional neural network structure is relatively simple.In order to achieve a faster and more accurate identification of crop pests and diseases than existing pest and disease detection systems.Provide guidance and help for the early warning of crop pests and diseases,as well as information management and management during the whole life cycle of crops.In conclusion,combined with pest detection algorithm based on neural network,and designs and implements a remote agricultural monitoring system with a mobile terminal.The system is fully functional and the detection of pests and diseases is accurate and stable.The storage and analysis of pest and disease information data will provide the data foundation for the prevention and early warning of crop pests and diseases in the future,and will provide analysis tools for the formulation of pest control programs,and will play a role in promoting the development of future agricultural modernization.The market prospect is broad.