Rice sheath blight disease identification method and system based on ShuffleNetV2-Unet

Abstract

The invention discloses a rice sheath blight disease identification method and system based on ShuffleNetV2-Unet, and the method comprises the following steps: S1, collecting a rice disease data set which comprises images with rice sheath blight disease marks; s2, establishing a ShuffleNetV2-Unet model, wherein the ShuffleNetV2-Unet model is used for identifying the disease marks of the rice sheath blight disease; s3, the ShuffleNetV2-Unet model is trained by means of the rice disease data set in the step S1, and the trained ShuffleNetV2-Unet model is obtained; and S4, the trained ShuffleNetV2-Unet model is utilized to carry out rice sheath blight disease identification on an input picture. According to the method, the rice sheath blight disease is identified through the ShuffleNetV2-Unet model, the disease marks of the rice sheath blight disease can be effectively identified in a complex environment, the detection speed can be kept while high detection precision is kept, meanwhile, the requirement for the detection environment for detecting the rice sheath blight disease is low, and the operation for detecting the rice sheath blight disease is convenient.

Description

technical field [0001] The invention relates to the fields of image processing and pattern recognition, and more particularly, to a method and system for rice sheath blight recognition based on ShuffleNetV2-Unet. Background technique [0002] Rice sheath blight is one of the main diseases of rice planting. The disease mainly occurs on the leaf sheaths and leaves of rice, and its markings are usually irregular moiré-like. When the disease occurs, the leaves will turn yellow and die, which seriously affects the planting of rice. and production. The traditional method of diagnosing rice diseases is to manually identify the diseases of rice. Farmers rely on their own accumulated experience, reading books or consulting professionals to determine the type of disease. This method is inefficient and difficult to achieve real-time disease detection. Monitoring and timely processing. Therefore, it is very urgent to design a method that can detect rice diseases in real time and deter...

AI Technical Summary

Problems solved by technology

[0003] At present, there are few practical applications of machine vision for the detection of rice sheath blight. The main reasons are as follows: 1) The detection environment is relatively complicated, and the disease pattern of sheath blight is irregular, and there are different degrees of moiré on the sheath and leaves. shape, its color and lesions are similar to those of dead leaves, so it is easy to confuse them

2) Early identification of rice sheath blight is difficult

3) Rice sheath blight disease patterns are complex, without fixed and single characteristics, and it is difficult to accurately identify the category

4) The traditional rice sheath blight recognition algorithm has high requirements for the detection environment and is difficult to apply in practice

[0010] The existing methods have realized the identification of rice sheath blight to varying degrees, but the main feature of rice sheath blight is that there are different degrees of moiré on the leaf sheath and leaves, which will appear densely distributed in the actual scene. The detection method has a good recognition effect on regular objects, but it can accurately identify irregular and densely distributed rice sheath blight. At the same time, the existing technology has extremely high requirements for the detection environment of rice sheath blight , in the actual detection environment, the distribution of rice sheath blight disease lines is messy, which is difficult to detect by traditional algorithms

Therefore, although the existing methods based on traditional image processing have achieved certain results, there is still a lot of room for improvement in terms of accuracy and recognition efficiency.

Images