Research And Implementation Of Image Detection System For Peach Diseases And Pests

China is a big country in planting peach trees,which are planted in all regions of China and bring huge economic benefits to China every year.With the increase of peach planting area,the occurrence of peach diseases and pests has become an important factor restricting peach quality and yield.It is very important to detect diseases and pests quickly and accurately in the early stage and control them in time.Traditional methods of disease and pest identification have problems such as strong subjectivity and difficulty in identification.With the rapid development of artificial intelligence technology,especially deep learning,it provides powerful technical support for automatic detection of peach diseases and pests.In this paper,using information technology such as image processing and artificial intelligence,the automatic detection of peach tree diseases and pests images is realized,and the image detection system of peach tree diseases and pests is researched and realized,which helps farmers to accurately detect and identify diseases and pests,and provides relevant prevention and control measures,providing strong support for farmers to increase production and income.The main research work of this paper is as follows:(1)Construct a dataset of images of peach tree diseases and pests.Images of peach tree diseases and pests were collected from peach orchards in three regions such as Qingdao,and the images were adjusted to a uniform size for manual labeling.Because the collection of diseases and pests is affected by factors such as weather and seasons,and the occurrence of diseases and pests is phased,the number of collected images cannot meet the needs of model training.Therefore,in order to enhance the generalization ability of the model,five methods of "cramming",translation,brightness adjustment,flipping and adding Gaussian noise were used for data augmentation in this paper.The peach tree disease image data set and the pest image data set were constructed,and the data set was adjusted to the format of Pascal VOC2007 to provide the standard format data for the training of the deep learning model in the next step.(2)Aiming at the problem of small size of peach tree diseases and pests,using target detection technology,a RFBNet algorithm based on Kmeans++ is proposed to construct image detection models of diseases and pests respectively.The dilated convolution layer in RFBNet has the characteristic of increasing the receptive field without losing the resolution and has a better detection effect on small targets.In the process of model training,the size of the prior frame will affect the detection effect,so in order to further improve the detection accuracy,Kmeans++ algorithm is introduced to optimize the prior frame in the network.Experimental results show that the proposed detection algorithm is better than SSD and RFBNet detection algorithms,and is more suitable for peach disease and pest detection.(3)In order to meet the needs of practical applications,this paper transplants the trained RFBNet detection model to the Android system,researches and implements the peach tree disease and pest image detection system.The system mainly includes image collection and preprocessing,disease and pest inquiry,disease and pest detection,and expert guidance functions.The system provides farmers with expert advice on peach tree management and accurate information on diseases and pests,allowing farmers to use this system to accurately determine the types of diseases and pests,and provide related prevention and control measures.