Identification of Cotton Root Rot by Multifeature Selection from Sentinel-2 Images Using Random Forest

MS-Net: a novel lightweight and precise model for plant disease identification

The rapid development of image processing technology and the improvement of computing power in recent years have made deep learning one of the main methods for plant disease identification. Currently, many neural network models have shown better performance in plant disease identification. Typically, the performance improvement of the model needs to be achieved by increasing the depth of the network. However, this also increases the computational complexity, memory requirements, and training time, which will be detrimental to the deployment of the model on mobile devices. To address this problem, a novel lightweight convolutional neural network has been proposed for plant disease detection. Skip connections are introduced into the conventional MobileNetV3 network to enrich the input features of the deep network, and the feature fusion weight parameters in the skip connections are optimized using an improved whale optimization algorithm to achieve higher classification accuracy. In addition, the bias loss substitutes the conventional cross-entropy loss to reduce the interference caused by redundant data during the learning process. The proposed model is pre-trained on the plant classification task dataset instead of using the classical ImageNet for pre-training, which further enhances the performance and robustness of the model. The constructed network achieved high performance with fewer parameters, reaching an accuracy of 99.8% on the PlantVillage dataset. Encouragingly, it also achieved a prediction accuracy of 97.8% on an apple leaf disease dataset with a complex outdoor background. The experimental results show that compared with existing advanced plant disease diagnosis models, the proposed model has fewer parameters, higher recognition accuracy, and lower complexity.

Construction of a Winter Wheat Comprehensive Growth Monitoring Index Based on a Fuzzy Degree Comprehensive Evaluation Model of Multispectral UAV Data

Realizing real-time and rapid monitoring of crop growth is crucial for providing an objective basis for agricultural production. To enhance the accuracy and comprehensiveness of monitoring winter wheat growth, comprehensive growth indicators are constructed using measurements of above-ground biomass, leaf chlorophyll content and water content of winter wheat taken on the ground. This construction is achieved through the utilization of the entropy weight method (EWM) and fuzzy comprehensive evaluation (FCE) model. Additionally, a correlation analysis is performed with the selected vegetation indexes (VIs). Then, using unmanned aerial vehicle (UAV) multispectral orthophotos to construct VIs and extract texture features (TFs), the aim is to explore the potential of combining the two as input variables to improve the accuracy of estimating the comprehensive growth indicators of winter wheat. Finally, we develop comprehensive growth indicator inversion models based on four machine learning algorithms: random forest (RF); partial least squares (PLS); extreme learning machine (ELM); and particle swarm optimization extreme learning machine (PSO-ELM), and the optimal model is selected by comparing the accuracy evaluation indexes of the model. The results show that: (1) The correlation among the comprehensive growth indicators (CGIs) constructed by EWM (CGIewm) and FCE (CGIfce) and VIs are all improved to different degrees compared with the single indicators, among which the correlation between CGIfce and most of the VIs is larger. (2) The inclusion of TFs has a positive impact on the performance of the comprehensive growth indicator inversion model. Specifically, the inversion model based on ELM exhibits the most significant improvement in accuracy. The coefficient of determination (R2) values of ELM-CGIewm and ELM- CGIfce increased by 20.83% and 20.37%, respectively. (3) The CGIfce inversion model constructed by VIs and TFs as input variables and based on the ELM algorithm is the best inversion model (ELM-CGIfce), with R2 reaching 0.65. Particle swarm optimization (PSO) is used to optimize the ELM-CGIfce (PSO-ELM-CGIfce), and the precision is significantly improved compared with that before optimization, with R2 reaching 0.84. The results of the study can provide a favorable reference for regional winter wheat growth monitoring.

Early detection of cotton verticillium wilt based on root magnetic resonance images

Verticillium wilt (VW) is often referred to as the cancer of cotton and it has a detrimental effect on cotton yield and quality. Since the root system is the first to be infested, it is feasible to detect VW by root analysis in the early stages of the disease. In recent years, with the update of computing equipment and the emergence of large-scale high-quality data sets, deep learning has achieved remarkable results in computer vision tasks. However, in some specific areas, such as cotton root MRI image task processing, it will bring some challenges. For example, the data imbalance problem (there is a serious imbalance between the cotton root and the background in the segmentation task) makes it difficult for existing algorithms to segment the target. In this paper, we proposed two new methods to solve these problems. The effectiveness of the algorithms was verified by experimental results. The results showed that the new segmentation model improved the Dice and mIoU by 46% and 44% compared with the original model. And this model could segment MRI images of rapeseed root cross-sections well with good robustness and scalability. The new classification model improved the accuracy by 34.9% over the original model. The recall score and F1 score increased by 59% and 42%, respectively. The results of this paper indicate that MRI and deep learning have the potential for non-destructive early detection of VW diseases in cotton.

Recognition of spider mite infestations in jujube trees based on spectral-spatial clustering of hyperspectral images from UAVs

Spider mite infestations are a serious hazard for jujube trees in China. The use of remote sensing technology to evaluate the health of jujube trees in large-scale intensive agricultural production is an effective means of agricultural control. Hyperspectral remote sensing has a higher spectral resolution and richer spectral information than conventional multispectral remote sensing, which improves the detection of crop pests and diseases. We used hyperspectral remote sensing data from jujube fields infested with spider mite in Hotan Prefecture, Xinjiang to evaluate their use in monitoring this important pest. We fused spectral and spatial information from the hyperspectral images and propose a method of recognizing spider mite infestations of jujube trees. Our method is based on the construction of spectral features, the fusion of spatial information and clustering of these spectral-spatial features. We evaluated the effect of different spectral-spatial features and different clustering methods on the recognition of spider mite in jujube trees. The experimental results showed that the overall accuracy of the method for the recognition of spider mites was >93% and the overall accuracy of the band clustering-density peak clustering model for the recognition of spider mite reached 96.13%. This method can be applied to the control of jujube spider mites in agricultural production.

Cotton aphid infestation monitoring using Sentinel-2 MSI imagery coupled with derivative of ratio spectroscopy and random forest algorithm

Aphids are one of the main pests of cotton and have been an important disaster limiting cotton yield. It is important to use satellite multispectral data to monitor the severity of cotton aphids in a timely and accurate manner on regional scale. Based on the combination of derivative of ratio spectra (DRS) and random forest (RF) algorithm, this study researched the quantitative monitoring model of cotton aphid severity based on Sentinel-2 data. First, the cotton area was extracted by using a supervised classification algorithm and the vegetation index threshold method. Then, the DRS algorithm was used to analyze the spectral characteristics of cotton aphids from three scales, and the Pearson correlation analysis algorithm was used to extract the bands significantly related to aphid infestation. Finally, the RF model was trained by ground sampling points and its accuracy was evaluated. The optimal model results were selected by the cross-validation method, and the accuracy was compared with the four classical classification algorithms. The results showed that (1) the canopy spectral reflectance curves at different grades of cotton aphid infestation were significantly different, with a significant positive correlation between cotton aphid grade and spectral reflectance in the visible band range and a negative correlation in the near-infrared band range; (2) The DRS algorithm could effectively remove the interference of the background endmember of satellite multispectral image pixels and enhance the aphid spectral features. The analysis results from three different scales and the evaluation results demonstrate the effectiveness of the algorithm in processing satellite multispectral data; (3) After the DRS processing, Sentinel-2 multispectral images could effectively classify the severity of cotton aphid infestation by the RF model with an overall classification accuracy of 80% and a kappa coefficient of 0.73. Compared with the results of four classical classification algorithms, the proposed algorithm has the best accuracy, which proves the superiority of RF. Based on satellite multispectral data, the DRS and RF can be combined to monitor the severity of cotton aphids on a regional scale, and the accuracy can meet the actual need.

Research hotspots and frontiers in agricultural multispectral technology: Bibliometrics and scientometrics analysis of the Web of Science

Multispectral technology has a wide range of applications in agriculture. By obtaining spectral information during crop production, key information such as growth, pests and diseases, fertilizer and pesticide application can be determined quickly, accurately and efficiently. The scientific analysis based on Web of Science aims to understand the research hotspots and areas of interest in the field of agricultural multispectral technology. The publications related to agricultural multispectral research in agriculture between 2002 and 2021 were selected as the research objects. The softwares of CiteSpace, VOSviewer, and Microsoft Excel were used to provide a comprehensive review of agricultural multispectral research in terms of research areas, institutions, influential journals, and core authors. Results of the analysis show that the number of publications increased each year, with the largest increase in 2019. Remote sensing, imaging technology, environmental science, and ecology are the most popular research directions. The journal Remote Sensing is one of the most popular publishers, showing a high publishing potential in multispectral research in agriculture. The institution with the most research literature and citations is the USDA. In terms of the number of papers, Mtanga is the author with the most published articles in recent years. Through keyword co-citation analysis, it is determined that the main research areas of this topic focus on remote sensing, crop classification, plant phenotypes and other research areas. The literature co-citation analysis indicates that the main research directions concentrate in vegetation index, satellite remote sensing applications and machine learning modeling. There is still a lot of room for development of multi-spectrum technology. Further development can be carried out in the areas of multi-device synergy, spectral fusion, airborne equipment improvement, and real-time image processing technology, which will cooperate with each other to further play the role of multi-spectrum in agriculture and promote the development of agriculture.

Combining Disease Mechanism and Machine Learning to Predict Wheat Fusarium Head Blight

Wheat Fusarium head blight (FHB) can be effectively controlled through prediction. To address the low accuracy and poor stability of model predictions of wheat FHB, a prediction method of wheat FHB that couples a logistic regression mechanism-based model and k-nearest neighbours (KNN) model is proposed in this paper. First, we selected predictive factors, including remote sensing-based and meteorological factors. Then, we quantitatively expressed the factor weights of the disease occurrence and development mechanisms in the disease prediction model by using a logistic model. Subsequently, we integrated the obtained factor weights into the predictive factors and input the predictive factors with weights into the KNN model to predict the incidence of wheat FHB. Finally, the accuracy and generalizability of the models were evaluated. Wheat fields in Changfeng, Dingyuan, Fengyuan, and Feidong counties, Anhui Province, where wheat FHB often occurs, were used as the study area. The incidences of wheat FHB on 29 April and 10 May 2021 were predicted. Compared with a model that did not consider disease mechanism, the accuracy of our model increased by approximately 13%. The overall accuracies of the models for the two dates were 0.88 and 0.92, and the F1 index was 0.86 and 0.94, respectively. The results show that the predictions made with the logistic-KNN model had higher accuracy and better stability than those made with the KNN model, thus achieving remote sensing-based high-precision prediction of wheat FHB.

Remote Sensing Monitoring of Winter Wheat Stripe Rust Based on mRMR-XGBoost Algorithm

For the problem of multi-dimensional feature redundancy in remote sensing detection of wheat stripe rust using reflectance spectrum and solar-induced chlorophyll fluorescence (SIF), a feature selection and disease index (DI) monitoring model combining mRMR and XGBoost algorithm was proposed in this study. Firstly, characteristic wavelengths selected by successive projections algorithm (SPA) were combined with the vegetation indices, trilateral parameters, and canopy SIF parameters to constitute the initial feature set. Then, the max-relevance and min-redundancy (mRMR) algorithm and correlation coefficient (CC) analysis were used to reduce the dimensionality of the initial feature set, respectively. Features selected by mRMR and CC were input as independent variables into the extreme gradient boosting regression (XGBoost) and gradient boosting regression tree (GBRT) to monitor the severity of stripe rust. The experimental results show that, compared with CC analysis, the monitoring accuracy of the features selected by mRMR in the XGBoost and GBRT models increased by 12% and 17% on average, respectively. Meanwhile, the mRMR-XGBoost model achieved the best monitoring accuracy (R2 = 0.8894, RMSE = 0.1135). The R2 between the measured DI and predicted DI of mRMR-XGBoost was improved by an average of 5%, 12%, and 22% compared with mRMR-GBRT, CC-XGBoost, and CC-GBRT models. These results suggested that XGBoost is more suitable for the remote sensing monitoring of wheat stripe rust, and mRMR has more advantages than the commonly used CC analysis in feature selection. Field survey data validation results also confirm that the mRMR-XGBoost algorithm has excellent monitoring applicability and scalability. The proposed model could provide a reference for data dimensionality reduction and crop disease index monitoring based on hyperspectral data.

Identification of Abandoned Jujube Fields Using Multi-Temporal High-Resolution Imagery and Machine Learning

The jujube industry plays a very important role in the agricultural industrial structure of Xinjiang, China. In recent years, the abandonment of jujube fields has gradually emerged. It is critical to inventory the abandoned land soon after it is generated to adjust agricultural production better and prevent the negative impacts from the abandonment (such as outbreaks of diseases, insect pests, and fires). High-resolution multi-temporal satellite remote sensing images can be used to identify subtle differences among crops and provide a good tool for solving this problem. In this research, both field-based and pixel-based classification approaches using field boundaries were used to estimate the percentage of abandoned jujube fields with multi-temporal high spatial resolution satellite images (Gaofen-1 and Gaofen-6) and the Random Forest algorithm. The results showed that both approaches produced good classification results and similar distributions of abandoned fields. The overall accuracy was 91.1% for the field-based classification and 90.0% for the pixel-based classification, and the Kappa was 0.866 and 0.848 for the respective classifications. The areas of abandoned land detected in the field-based and pixel-based classification maps were 806.09 ha and 828.21 ha, respectively, accounting for 8.97% and 9.11% of the study area. In addition, feature importance evaluations of the two approaches showed that the overall importance of texture features was higher than that of vegetation indices and that 31 October and 10 November were important dates for abandoned land detection. The methodology proposed in this study will be useful for identifying abandoned jujube fields and have the potential for large-scale application.