Research On Tea Pest Prevention And Control System Based On UAV

As new agricultural machinery,eppo UAV has the advantages of high operating efficiency,good maneuverability,and high safety and stability.At present,it has been applied more and more widely.In order to ensure the precision of pesticide spraying,it is particularly important to accurately locate diseases and insect pests in the early stage.Therefore,it has great significance to establish a fast and convenient and efficient model,which identify disease and insect pest areas.In this paper,proposed a tea disease and insect pest image detection model based on Faster R-CNN,so as to realize accurate location of diseases and insect pests,providing technical support for the fine operation of plant protection UAV.The main works are in the following:1.A four-rotor plant protection UAV is designed based on the control system of Pixhawk,which can steadily perform autonomous cruise,intelligent obstacle avoidance,image acquisition,intelligent spraying,and other tasks.2.Firstly,the median filter algorithm is used to reduce the noise of the collected images containing noise information.Then the histogram equalization algorithm is used to enhance the image of tea leaf diseases and pests.Based on the deep convolution neural network's strong ability in image feature representation and learning ability in model generalization,a Faster R-CNN based image detection algorithm for tea diseases and insect pests was proposed.3.A method based on Faster R-CNN is proposed to train the image detection Model of tea plant diseases and insect pests,so as to train a target detection network with high robustness and feature representation ability,and the experiment was compared with other traditional target detection algorithms such as Deformable Part Model,R-CNN and Faster R-CNN.The results showed that Faster R-CNN algorithm was the most optimal and could achieve 98.7% accuracy.The paper has 52 pictures,10 tables and 61 references.