Estimation of Leaf Nitrogen Content in Wheat Based on Fusion of Spectral Features and Deep Features from Near Infrared Hyperspectral Imagery

The prediction model of nitrogen nutrition in cotton canopy leaves based on hyperspectral visible-near infrared band feature fusion

Hyperspectral remote sensing technology is becoming increasingly popular in various fields due to its ability to provide detailed information about crop growth and nutritional status. The use of hyperspectral technology to predict SPAD (Soil and Plant Analyzer Development) values during cotton growth and adopt precise fertilization management measures is crucial for achieving high yield and fertilizer efficiency. To detect the nitrogen nutrition in cotton canopy leaves quickly, a non-destructive nitrogen nutrition retrieval model was proposed based on the spectral fusion features of the cotton canopy. The hyperspectral vegetation index and multifractal features were fused to predict the SPAD value and identify the amount of fertilizer applied at different levels. The random decision forest algorithm was used as the model predictor and classifier. A method was introduced which was widely used in the fields of finance and stocks (MF-DFA) into the field of agriculture to extract fractal features of cotton spectral reflectance. Comparing the fusion feature with multi-fractal feature and vegetation index, the results showed that the fusion feature parameters had higher accuracy and better stability than using a single feature or feature combination. The R2 was as high as 0.8363, and the RMSE was 1.8767%. Our intelligent model provides a new idea for detecting nitrogen nutrition in cotton canopy leaves rapidly.

Drought stress identification of tomato plant using multi-features of hyperspectral imaging and subsample fusion

Drought stress (DS) is one of the most frequently occurring stresses in tomato plants. Detecting tomato plant DS is vital for optimizing irrigation and improving fruit quality. In this study, a DS identification method using the multi-features of hyperspectral imaging (HSI) and subsample fusion was proposed. First, the HSI images were measured under imaging condition with supplemental blue lights, and the reflectance spectra were extracted from the HSI images of young and mature leaves at different DS levels (well-watered, reduced-watered, and deficient-watered treatment). The effective wavelengths (EWs) were screened by the genetic algorithm. Second, the reference image was determined by ReliefF, and the first four reflectance images of EWs that are weakly correlated with the reference image and mutually irrelevant were obtained using Pearson's correlation analysis. The reflectance image set (RIS) was determined by evaluating the superposition effect of reflectance images on identification. The spectra of EWs and the image features extracted from the RIS by LeNet-5 were adopted to construct DS identification models based on support vector machine (SVM), random forest, and dense convolutional network. Third, the subsample fusion integrating the spectra and image features of young and mature leaves was used to improve the identification further. The results showed that supplemental blue lights can effectively remove the high-frequency noise and obtain high-quality HSI images. The positive effect of the combination of spectra of EWs and image features for DS identification proved that RIS contains feature information pointing to DS. Global optimal classification performance was achieved by SVM and subsample fusion, with a classification accuracy of 95.90% and 95.78% for calibration and prediction sets, respectively. Overall, the proposed method can provide an accurate and reliable analysis for tomato plant DS and is hoped to be applied to other crop stresses.

Analysis of the Application Efficiency of TensorFlow and PyTorch in Convolutional Neural Network

In this paper, we present an analysis of important aspects that arise during the development of neural network applications. Our aim is to determine if the choice of library can impact the system's overall performance, either during training or design, and to extract a set of criteria that could be used to highlight the advantages and disadvantages of each library under consideration. To do so, we first extracted the previously mentioned aspects by comparing two of the most popular neural network libraries-PyTorch and TensorFlow-and then we performed an analysis on the obtained results, with the intent of determining if our initial hypothesis was correct. In the end, the results of the analysis are gathered, and an overall picture of what tasks are better suited for what library is presented.

Applications of hyperspectral imaging in plant phenotyping

Our ability to interrogate and manipulate the genome far exceeds our capacity to measure the effects of genetic changes on plant traits. Much effort has been made recently by the plant science research community to address this imbalance. The responses of plants to environmental conditions can now be defined using a variety of imaging approaches. Hyperspectral imaging (HSI) has emerged as a promising approach to measure traits using a wide range of wavebands simultaneously in 3D to capture information in lab, glasshouse, or field settings. HSI has been applied to define abiotic, biotic, and quality traits for optimisation of crop management.

Hyperspectral Characteristics of an Individual Leaf of Wheat Grown under Nitrogen Gradient

Since the application of hyperspectral technology to agriculture, many scientists have been conducting studies to apply the technology in crop diagnosis. However, due to the properties of optical devices, the reflectances obtained according to the image acquisition conditions are different. Nevertheless, there is no optimized method for minimizing such technical errors in applying hyperspectral imaging. Therefore, this study was conducted to find the appropriate image acquisition conditions that reflect the growth status of wheat grown under different nitrogen fertilization regimes. The experiment plots were comprised of six plots with various N application levels of 145.6 kg N ha^-1 (N1), 109.2 kg N ha^-1 (N2), 91.0 kg N ha^-1 (N3), 72.8 kg N ha^-1 (N4), 54.6 kg N ha^-1 (N5), and 36.4 kg N ha^-1 (N6). Hyperspectral image acquisitions were performed at different shooting angles of 105° and 125° from the surface, and spike, flag leaf, and the second uppermost leaf were divided into five parts from apex to base when analyzing the images. The growth analysis conducted at heading showed that the N6 was 85.6% in the plant height, 44.1% in LAI, and 64.9% in SPAD as compared to N1. The nitrogen content in the leaf decreased by 55.2% compared to N1 and the quantity was 44.9% in N6 compared to N1. Based on the vegetation indices obtained from hyperspectral reflectances at the heading stage, the spike was not suitable for analysis. In the case of the flag leaf and the 2nd uppermost leaf, the vegetation indices from spectral data taken at 105 degrees were more appropriate for acquiring imaging data by clearly dividing the effects of fertilization level. The results of the regional variation in a leaf showed that the region of interest (ROI), which is close to the apex of the flag leaf and the base of the second uppermost leaf, has a high coefficient of determination between the fertilization levels and the vegetation indices, which effectively reflected the status of wheat.

Accurate Wheat Lodging Extraction from Multi-Channel UAV Images Using a Lightweight Network Model

The extraction of wheat lodging is of great significance to post-disaster agricultural production management, disaster assessment and insurance subsidies. At present, the recognition of lodging wheat in the actual complex field environment still has low accuracy and poor real-time performance. To overcome this gap, first, four-channel fusion images, including RGB and DSM (digital surface model), as well as RGB and ExG (excess green), were constructed based on the RGB image acquired from unmanned aerial vehicle (UAV). Second, a Mobile U-Net model that combined a lightweight neural network with a depthwise separable convolution and U-Net model was proposed. Finally, three data sets (RGB, RGB + DSM and RGB + ExG) were used to train, verify, test and evaluate the proposed model. The results of the experiment showed that the overall accuracy of lodging recognition based on RGB + DSM reached 88.99%, which is 11.8% higher than that of original RGB and 6.2% higher than that of RGB + ExG. In addition, our proposed model was superior to typical deep learning frameworks in terms of model parameters, processing speed and segmentation accuracy. The optimized Mobile U-Net model reached 9.49 million parameters, which was 27.3% and 33.3% faster than the FCN and U-Net models, respectively. Furthermore, for RGB + DSM wheat lodging extraction, the overall accuracy of Mobile U-Net was improved by 24.3% and 15.3% compared with FCN and U-Net, respectively. Therefore, the Mobile U-Net model using RGB + DSM could extract wheat lodging with higher accuracy, fewer parameters and stronger robustness.

Machine Learning in Agriculture: A Comprehensive Updated Review

The digital transformation of agriculture has evolved various aspects of management into artificial intelligent systems for the sake of making value from the ever-increasing data originated from numerous sources. A subset of artificial intelligence, namely machine learning, has a considerable potential to handle numerous challenges in the establishment of knowledge-based farming systems. The present study aims at shedding light on machine learning in agriculture by thoroughly reviewing the recent scholarly literature based on keywords' combinations of "machine learning" along with "crop management", "water management", "soil management", and "livestock management", and in accordance with PRISMA guidelines. Only journal papers were considered eligible that were published within 2018-2020. The results indicated that this topic pertains to different disciplines that favour convergence research at the international level. Furthermore, crop management was observed to be at the centre of attention. A plethora of machine learning algorithms were used, with those belonging to Artificial Neural Networks being more efficient. In addition, maize and wheat as well as cattle and sheep were the most investigated crops and animals, respectively. Finally, a variety of sensors, attached on satellites and unmanned ground and aerial vehicles, have been utilized as a means of getting reliable input data for the data analyses. It is anticipated that this study will constitute a beneficial guide to all stakeholders towards enhancing awareness of the potential advantages of using machine learning in agriculture and contributing to a more systematic research on this topic.

IndoorPlant: A Model for Intelligent Services in Indoor Agriculture Based on Context Histories

The application of ubiquitous computing has increased in recent years, especially due to the development of technologies such as mobile computing, more accurate sensors, and specific protocols for the Internet of Things (IoT). One of the trends in this area of research is the use of context awareness. In agriculture, the context involves the environment, for example, the conditions found inside a greenhouse. Recently, a series of studies have proposed the use of sensors to monitor production and/or the use of cameras to obtain information about cultivation, providing data, reminders, and alerts to farmers. This article proposes a computational model for indoor agriculture called IndoorPlant. The model uses the analysis of context histories to provide intelligent generic services, such as predicting productivity, indicating problems that cultivation may suffer, and giving suggestions for improvements in greenhouse parameters. IndoorPlant was tested in three scenarios of the daily life of farmers with hydroponic production data that were obtained during seven months of cultivation of radicchio, lettuce, and arugula. Finally, the article presents the results obtained through intelligent services that use context histories. The scenarios used services to recommend improvements in cultivation, profiles and, finally, prediction of the cultivation time of radicchio, lettuce, and arugula using the partial least squares (PLS) regression technique. The prediction results were relevant since the following values were obtained: 0.96 (R², coefficient of determination), 1.06 (RMSEC, square root of the mean square error of calibration), and 1.94 (RMSECV, square root of the mean square error of cross validation) for radicchio; 0.95 (R²), 1.37 (RMSEC), and 3.31 (RMSECV) for lettuce; 0.93 (R²), 1.10 (RMSEC), and 1.89 (RMSECV) for arugula. Eight farmers with different functions on the farm filled out a survey based on the technology acceptance model (TAM). The results showed 92% acceptance regarding utility and 98% acceptance for ease of use.