Design And Implementation Of Deep Learning Based Plant Leaf Disease Identification System

China is a large agricultural country,and ensuring the normal supply of agricultural products is very important.However,plant damage caused by severe weather,plant diseases,and insect pests can lead to a large number of plant wilt and crop yield reduction every year.As the country has been implementing the policy of returning farmland to forests,China’s cultivated land cannot be significantly increased in a short period.Therefore,on the existing cultivated land,it is particularly important to ensure normal output and increase production and income.Therefore,protecting plants from pests and diseases and improving crop yields have always been major issues that our country is committed to solving.Based on the above background,this paper carried out research on plant leaf disease recognition based on deep learning.It includes the following aspects:First,this paper uses deep learning and convolutional neural network to judge whether plant leaves are diseased and how diseased they are,studies the key algorithms of plant leaf disease image recognition system based on deep learning,and uses Yolo v5 model to identify soybean leaves and leaf disease spots.The experimental results show that compared with Yolo V3,the improved Yolo v5 model has higher recognition accuracy and faster recognition rate in the application process of plant leaf disease recognition,and has more practical value.Secondly,on the basis of the Yolo v5 model,further research was carried out,mainly in two aspects of improvement: in terms of speed,Shuffle Net V2 was used to replace some original network layers of the Yolo v5 model,and speed was improved with a relatively small loss of accuracy;In terms of accuracy,this paper uses Varifocal Loss Loss function to replace Focal Loss of Yolo v5 model to improve the precision of small target recognition.Finally,based on the improved algorithm,this paper designed and implemented a plant leaf disease image recognition system and mobile APP based on deep learning model.The system and mobile APP can realize model management,data management,disease recognition and other functions,and provide a visual display interface.The built-in deep learning model not only automatically identifies plant leaf disease images,but also provides users with the function of plot management.Through machine learning and statistical analysis knowledge,users can show the crop disease analysis of the plots set by users.