Rice Maturity Detection System Based On Drone And Convolutional Neural Networ

With the development of our country’s economy and the improvement of farmers’ living standards,many rice brands with characteristics have emerged.Their varieties,taste and cooking methods are similar,bringing people different eating experiences.People have more choices and requirements for the selection of rice.Bright color,good taste and long storage cycle are the primary judgment basis for high-quality rice,and maturity is the key factor affecting the quality of rice.At the earliest,rice maturity was judged by human visual observation combined with traditional planting experience.With the development of The Times,rice maturity detection uses chemical methods to analyze the content changes of starch,protein,water content of rice kernels to determine the maturity of rice ears,and accurately measure the maturity of rice.However,relying on traditional experience to determine rice maturity has limitations.Although chemical analysis is accurate,it is time-consuming,and ordinary users are limited by scientific knowledge and experimental conditions,so they cannot quickly obtain results.There is no effective and rapid method for detecting rice maturity in a large area of farmland.Unmanned aerial Vehicle(UAV)has the characteristics of flexibility,speed and efficiency.However,its flight is limited by its own battery and environmental obstacles,so its endurance is short and its shooting efficiency is low.This paper uses the improved particle swarm optimization algorithm and bat algorithm to optimize the flight path of the UAV in the rice field,so that it can save energy and take pictures of rice in the rice field efficiently.This paper designed a set of rice With the development of our country’s economy and the improvement of farmers’ living standards,many rice brands with characteristics have emerged.Their varieties,taste and cooking methods are similar,bringing people different eating experiences.People have more choices and requirements for the selection of rice.Bright color,good taste and long storage cycle are the primary judgment basis for high-quality rice,and maturity is the key factor affecting the quality of rice.At the earliest,rice maturity was judged by human visual obsermaturity detection system,and applied the convolutional neural network algorithm to the detection of rice maturity.Compared with the traditional experience judgment and chemical analysis,this system can quickly detect the maturity of rice,and has the characteristics of easy to use and simple operation.Firstly,this paper studies the research status of agricultural intelligence at home and abroad,analyzes the actual needs of the maturity detection system,and on this basis,puts forward the design scheme of rice maturity detection system in line with people’s living habits,and designs the function modules of the system.Secondly,the key technologies and theoretical knowledge used in the rice maturity detection system are deeply studied,which mainly include neural network technology,particle swarm optimization algorithm,bat algorithm,particle swarm optimization combined with bat algorithm,Spring Boot,Html5,Vue and Android Application.The particle swarm optimization combined with bat algorithm is applied to the path optimization of UAV to design an efficient,safe and low-cost flight path of UAV.The convolutional neural network algorithm is applied to the rice maturity detection module to realize a fast and accurate rice maturity identification method.The system is divided into system management platform based on the browser and APP early warning terminal based on the Android platform.The management platform includes the implementation of functional modules such as administrator login,management of ordinary users,image management,maturity detection result management,system log management,etc.The APP early warning terminal is designed with modules such as user,image upload,maturity detection,etc.Users can upload images of rice at any time to detect the maturity of rice.Finally,a system test platform was built,and matlab software was used to simulate whether UAV path planning reached expectations.Whether the management platform and the APP early warning terminal can be used normally,and the actual detection results of the rice maturity detection module.After testing,the particle swarm optimization combined with bat algorithm is superior to bat algorithm and particle swarm optimization algorithm in three aspects of cost fitness,algorithm execution time and algorithm self-fitness value.The planned path has low cost,short flight distance and short algorithm execution time.Rice maturity detection results test,100 test rice pictures,90 actual detection results,detection success rate of 90%.The system has completed various functions,which verifies the feasibility of the system.