Research On Maize Leaf Disease And Pest Identification System Based On Improved YOLOv5s

Corn has short production cycle,high yield,strong drought resistance,cold resistance,good environmental adaptability and other characteristics,so it is widely planted.China has the largest corn planting and harvesting area in the world,but in recent years,corn diseases and insect pests have become more and more serious,and there are many kinds of diseases and insect pests,which result in sharp reduction of corn yield and huge economic losses.Therefore,it is very important to identify and solve the problem of maize pests and diseases in time.In the early stage,the identification of maize leaf diseases and insect pests is more dependent on artificial subjective judgment,but affected by personal subjective factors,it is easy to appear wrong judgment and inspection omission,which cannot guarantee the unified identification accuracy.In addition,artificial inspection will also occur in the inspection is not timely,and the problem of missing the best time to control.In this thesis,the image of maize leaf disease and insect pests taken in the field is taken as the research object,the automatic detection model of maize leaf disease and insect pests is designed by using the object detection technology based on deep learning,and the automatic identification system of maize leaf disease and insect pests is realized.The research work is as follows:Firstly,the data set of maize leaf pests and diseases was constructed.The collection method is open source data obtained through field photography and kaggle platform.The images cover six most common maize leaf diseases and insect pests,namely,maize large spot,maize small spot,maize grey spot,maize rust,armyworm eggs and armyworm larvae.The obtained maize leaf disease and insect pests images are normalized,marked manually,data enhanced,sorted out and summarized to construct maize leaf disease and insect pests image data set required by the document for the subsequent experimental stage.Secondly,based on YOLOv5 s model,the following improvements were made to the model:(1)K-means++ was used instead of K-means to extract prior boxes;(2)Improved C3 module;(3)Context feature fusion is carried out on backbone to further detect maize leaf diseases and insect pests.By comparing the performance with other detection models,it is proved that the above improvements can improve the performance of the model and effectively detect the image information of maize leaf pests and diseases.Finally,a maize leaf pest identification system based on the improved YOLOv5 s was designed and implemented.The system has complete functions,is easy to use and accurate detection,can effectively identify maize leaf diseases and pests,and can provide better user experience,convenient operation,is conducive to the identification and management of diseases and pests in the agricultural field.