Wheat rust identification method based on transfer learning and sharpness perception minimization

Abstract

The invention provides a wheat rust recognition method based on transfer learning and sharpness perception minimization, and the method is characterized in that the method comprises the following steps: S1, obtaining wheat rust pictures in different disease periods, and grading the wheat rust pictures according to the severity of diseases; s2, performing preprocessing and data enhancement on the wheat rust pictures to obtain a wheat rust training set; s3, pre-training the lightweight model by using the ImageNet large-scale picture data set to obtain a pre-trained model, and storing parameters of the pre-trained model; and S4, on the basis of the pre-training model, training the pre-training model by utilizing the wheat rust training set, and finely adjusting the pre-training model by adopting a label smooth loss function in combination with a sharpness perception minimization method, thereby improving the generalization ability of the pre-training model. By introducing the transfer learning method, the recognition precision of the lightweight model is effectively improved, and the risk of model overfitting can be prevented in combination with the sharpness perception minimization method.

Description

technical field [0001] The invention relates to the field of crop disease classification, in particular to a wheat rust identification method based on migration learning and sharpness perception minimization. Background technique [0002] Wheat is one of the most important food crops, and its total output ranks second among all food crops. Wheat yield is affected by pests and diseases. Wheat rust is one of the more common diseases. When the disease occurs, it can generally reduce wheat yield by 5 to 15%, and it can reach more than 50% in severe cases. [0003] The traditional prevention and control of wheat rust mainly relies on experienced farmers to detect and treat the diseased wheat in the early stage of the disease. This requires farmers to have higher knowledge of disease prevention and control, and it is difficult to ensure the timeliness of disease detection. At present, with the development of automation, artificial intelligence and other technologies, it is a fea...

AI Technical Summary

Problems solved by technology

However, due to the hardware limitations of the device, it is usually necessary to load a relatively lightweight model on the UAV

Compared with the traditional deep model, the lightweight model requires less storage space and lower computing power requirements, but it also leads to loss of accuracy and possible overfitting problems

Images