Crop leaf disease identification method based on deep convolutional neural network

Abstract

The present invention discloses a crop leaf disease identification method based on a deep convolutional neural network, belonging to the field of image processing and machine learning technology. Themethod comprises the steps of: performing extension of an original image database to obtain an extended image database; constructing a deep convolutional neural network, and employing the extended image database to perform training of the deep convolutional neural network to obtain a crop disease identification model; inputting leaf images of crops to be identified into the crop disease identification model to obtain feature vectors, and performing image classification of the feature vectors; and obtaining types of diseases of the crops to be identified according to category labels of the crops to be identified and an image classification result. The method provided by the invention can improve the accuracy rate of the crop disease identification model, and can improve the applicability ofthe crop disease identification model.

Description

technical field [0001] The present disclosure relates to the technical fields of image processing and machine learning, and in particular, relates to a method for identifying crop leaf diseases based on a deep convolutional neural network. Background technique [0002] Crop diseases will not only affect the growth of crops, but also affect the yield and quality of crops; therefore, timely detection of crop disease occurrence and type identification is of great significance for the prevention and control of crop diseases. Among them, the key to effective prevention and control of crop diseases is how to quickly diagnose the disease type and development status, and formulate effective control methods and dosage according to the disease type and development status. [0003] In the traditional disease recognition method based on crop leaf image, it can be realized by extracting the color, shape and texture of the diseased leaf image. Among them, the extraction and selection of ...

AI Technical Summary

Problems solved by technology

For example, Arivazhagan et al. proposed a crop disease leaf image detection and recognition method; Wang Xianfeng et al. extracted the color, shape, texture and other features of the diseased leaf image lesion, combined with the environmental information of crop growth, and used discriminant analysis to identify cucumbers. Lesion category; Zhang et al. first carried out the spot segmentation of the leaf image, then extracted the color, shape and texture features of the lesion, and then identified five kinds of corn leaves through the K-nearest neighbor classification algorithm; the above algorithm is mainly by extracting The specific characteristics of the diseased leaf image, combined with the traditional classification method to identify the disease, although a good recognition effect has been achieved, but due to the complexity of the crop diseased leaf image, the extracted specific features cannot completely and well represent the crop. Lesion features of diseased leaf images have certain limitations. For example, some features are only suitable for representing disease spots with clear textures, while others are only suitable for representing disease spots with clear outlines; further, different features are not suitable for identifying Effects can also have adverse effects, so selecting a set of features suitable for disease type identification requires a lot of experimental research and experience summarization, which is a time-consuming and difficult problem

Images