A Deep-Learning-Based Approach for Wheat Yellow Rust Disease Recognition from Unmanned Aerial Vehicle Images.

Qian Pan,Jingwen Yan,Pingbo Wu,Maofang Gao,Shilei Li

doi:10.3390/s21196540

Abstract

Yellow rust is a disease with a wide range that causes great damage to wheat. The traditional method of manually identifying wheat yellow rust is very inefficient. To improve this situation, this study proposed a deep-learning-based method for identifying wheat yellow rust from unmanned aerial vehicle (UAV) images. The method was based on the pyramid scene parsing network (PSPNet) semantic segmentation model to classify healthy wheat, yellow rust wheat, and bare soil in small-scale UAV images, and to investigate the spatial generalization of the model. In addition, it was proposed to use the high-accuracy classification results of traditional algorithms as weak samples for wheat yellow rust identification. The recognition accuracy of the PSPNet model in this study reached 98%. On this basis, this study used the trained semantic segmentation model to recognize another wheat field. The results showed that the method had certain generalization ability, and its accuracy reached 98%. In addition, the high-accuracy classification result of a support vector machine was used as a weak label by weak supervision, which better solved the labeling problem of large-size images, and the final recognition accuracy reached 94%. Therefore, the present study method facilitated timely control measures to reduce economic losses.

Highlights

Wheat yellow rust is characterized by wide distribution, rapid spread, and a large damage area in wheat production in China
This study compared the recognition effects of several categories using support vector machine (SVM), random forest (RF), BPNN, FCN, U-Net, and pyramid scene parsing network (PSPNet) models on the unmanned aerial vehicle (UAV) image dataset acquired from the Xigolou Village plot, investigated the spatial generalization of the PSPNet model with the best recognition effect on the UAV images from the Dahuai Village plot, and solved the sample labeling problem based on the weak-sample learning approach
The effectiveness of the PSPNet model is discussed in terms of model structure; the robustness of the PSPNet model is discussed based on the better recognition results of the model; and the efficiency of the weakly supervised learning method is discussed in the case for which manual labeling of large region samples was not feasible

Summary

Introduction

Wheat yellow rust is characterized by wide distribution, rapid spread, and a large damage area in wheat production in China. With the rapid development of remote-sensing (RS) technology, aerial remote sensing is gradually becoming an important complementary approach This approach of using an unmanned aerial vehicle (UAV) to obtain RS data for relevant research is gradually becoming widely used for crop disease identification, and makes it possible to identify crop diseases at a coarse scale. Zhang et al [7] established a method to quantitatively characterize the shape features around specific absorption locations in the spectrum by artificially inoculating wheat with powdery mildew fungus after nutrient stress treatment, based on the differences in canopy spectral reflectance of wheat with different disease conditions. The number of NIR spectra of several crop diseases varied widely, causing the identification accuracy to vary widely, so the effectiveness of the discriminant partial least squares and SVM algorithms in the identification task could not be fully determined. Traditional model inversion algorithms are not effective enough to meet the current needs of identifying crop diseases, and the emergence of methods that can efficiently identify crop diseases is required

Methods

Results

Discussion

Conclusion