Identification Of Soybean Diseases And Pests And Varieties Based On Remote Sensing Imaging Technology

Accurate identification of diseases and pests and seed varieties are great significance for improving crop yield and quality,reducing economic losses,and ensuring food security.Soybean diseases and pests' digital images and soybean seed hyperspectral images were obtained,deep learning and spectral analysis techniques were used to classify and identify them,and the identification models of soybean diseases and pests and soybean seed varieties were constructed.The main results are as follows:1.Images were collected of 14 kinds of soybean diseases and pests in different environment by mobile phone camera.A total of 18016 images were obtained after data enhancement and the images were divided into a training set,validation set and test set with a 7:2:1 ratio.Pretrained models such as AlexNet were used for transfer training after fine-tuned,and convolutional neural network recognition models for Soybean diseases and pests were established.The results showed that the eight models all achieved 88% accuracy in the validation set,and DenseNet-201,Xception and NASNet-Large had the higher validation accuracy of over 99%.NASNet-Large model was used to identified 1802 images in the test sets and obtained 14 misjudgments with a test error rate of only 0.78%.2.1200 seeds of ten soybean varieties were selected,hyperspectral images of both the front and the back of the seeds were collected,and the reflectance of soybean was derived from the hyperspectral images.Savitzky-Golay smoothing(SG),first derivative(FD),standard normal variate(SNV),fast Fourier transform(FFT),Hilbert transform(HT),and multiplicative scatter correction(MSC)spectral reflectance pretreatment methods were used.Then,the feature wavelengths and feature information of the pretreated spectral reflectance data were extracted using competitive adaptive reweighted sampling(CARS),the successive projections algorithm(SPA),and principal component analysis(PCA).Finally,5 classifiers,Bayes,support vector machine(SVM),k-nearest neighbor(KNN),ensemble learning(EL),and artificial neural network(ANN),were used to identify seed varieties.The results showed that MSC-CARS-EL had the highest accuracy among the 90 combinations,with training set,test set,and 5-fold cross-validation accuracies of 100%,100%,and 99.8%,respectively.Moreover,the contribution of spectral pretreatment to discrimination accuracy was higher than those of feature extraction and classifier selection.Pretreatment determined the range of the identification accuracy,feature-selective methods and classifiers only changed within this range.3.The wavelength information of the third-dimension of 2400 soybean hyperspectral images was extracted and retained the first three principal components.Then the images were added blank pixels,resize,and rotated.A total of 9600 images were obtained after data augmentation,and the images were divided into a training set,validation set,and test set with a 3:1:1 ratio.Pretrained models(AlexNet,ResNet-18,Xception,Inception-V3,DenseNet-201,and NASNet-Large)after fine-tuning were used for transfer training.The optimal CNN model for soybean seed variety identification was selected.Furthermore,the traditional machine learning models for soybean seed variety identification were established by using reflectance as input.The results show that the six models all achieved 91% accuracy in the validation set and achieved accuracy values of 90.6%,94.5%,95.4%,95.6%,96.8%,and 97.2%,respectively,in the test set.Both the soybean diseases and pests identification model and the soybean seed variety identification model achieved high recognition accuracy and strong stability.It accumulates some experience for the application of remote sensing technology in the agricultural field and lay a foundation for the development of precision agriculture and smart agriculture.