Rice Precision Agriculture Information System Based On Artificial Intelligence

Nowadays,environmental pollution,energy exhaustion and the reduction of cultivated land are all affecting agricultural production.To feed China’s 1.3billion population,it is necessary to increase agricultural grain production.This paper presents a rice precision agriculture information system based on artificial intelligence.The system can acquire real-time information of rice farmland and monitor the growth environment of rice intelligently so as to take timely measures to improve rice yield and save manpower and material resources.In this paper,the overall design of this system adopts four-tier Internet of Things structure,and integrates the image processing and analysis technology of artificial intelligence to realize the function of collecting and analyzing data from farmland.Data acquisition function is accomplished by different types of sensors.The collected information can be divided into two categories: meteorological information and image and GPS information.The collected data are transmitted to the cloud storage platform through two processes: farmland internal transmission and Internet transmission.The data transmission in farmland depends on the field data transmission module,which is equipped with wireless receiving module,and can receive the wireless signals from meteorological sensors.The FTP server is built on the field data transmission module,and the received meteorological data is stored in classified way.The network data transmission module is used as the FTP client.The remote transmission of data in the field is realized between the two modules.Internet transmission is realized through the network data transmission module,which has a 4G unit,can access the 4G network and upload data to the cloud storage platform.In any terminal equipped with Cool Object Link Client,access rights can be obtained through account number and password to view data and truly realize agricultural data sharing.Data analysis includes two steps: image processing and classification.Image data will be segmented and extracted by OpenCV,and then classified by convolutional neural network classifier to get the types of rice pests and diseases.In PC client,the above meteorological data,GPS data,image data and classification results are systematically displayed on the software interface.User login client software system can view the data collected by sensors and the analysis results and other information.After the actual test,the network data transmission module can upload all data classifications normally,download these data on the PC terminal,and the image data processing and classification process is normal.However,due to the unstable classification results of a single classifier,the accuracy rate can not meet the requirements.Finally,the idea of stochastic forest algorithm is introduced to obtain the final results based on the statistics of the classification results of multiple models,which makes the classification accuracy rate 96.32%.Finally,the working process of the client application is tested,and the functions of data call and data display are normal.