Search for: All records

Over the last decade, the use of unmanned aerial vehicles (UAVs) for plant phenotyping and field crop monitoring has significantly evolved and expanded. These technologies have been particularly valuable for monitoring crop growth and health and for managing abiotic and biotic stresses such as drought, fertilization deficiencies, disease, and bioaggressors. This paper provides a comprehensive review of the progress in UAV‐based plant phenotyping, with a focus on the current use and application of drone technology to gain information on plant growth, development, adaptation, and yield. We reviewed over 200 research articles and discuss the best tools and methodologies for different research purposes, the challenges that need to be overcome, and the major research gaps that remain. First, the review offers a critical focus on elucidating the distinct characteristics of UAV platforms, highlighting the diverse sensor technologies employed and shedding light on the nuances of UAV data acquisition and processing methodologies. Second, it presents a comprehensive analysis of the multiple applications of UAVs in field phenotyping, underscoring the transformative potential of integrating machine learning techniques for plant analysis. Third, it delves into the realm of machine learning applications for plant phenotyping, emphasizing its role in enhancing data analysis and interpretation. Furthermore, the paper extensively examines the open issues and research challenges within the domain, addressing the complexities and limitations faced during data acquisition, processing, and interpretation. Finally, it outlines the future trends and emerging technologies in the field of UAV‐based plant phenotyping, paving the way for innovative advancements and methodologies. 

Assessing soil organic carbon (SOC) stocks is crucial for understanding the carbon sequestration potential of agroecosystems and for mitigating climate change. This study presents a novel method for assessing SOC and mineral content at various soil depths in sorghum crops using hyperspectral remote sensing. Conducted at Planthaven Farms, MO, the research encompassed ten genotypes across 30 plots, yielding 180 soil samples from six depth intervals (0–150 cm) of bare soil. Chemical analyses determined the SOC and mineral levels, which were then compared to spectral data from HySpex indoor sensors. We utilized time-frequency analysis methods, including discrete wavelet transformation (DWT), continuous wavelet transformation (CWT), and frame transformation along with traditional spectral transformations, specifically fractional derivatives and continuum removal. The analysis revealed the shortwave infrared (SWIR) region, particularly the 1800–2000 nm range, as having the strongest correlations with SOC content (with R2 exceeding 0.8). The visible near-infrared (VNIR) region also provided valuable insights. Models incorporating CWT achieved high accuracy (test R2 exceeding 0.9), while frame transformation achieved strong accuracy (test R2 between 0.7 and 0.8) with fewer features. The random forest regressor (RFR) proved to be most robust, demonstrating superior accuracy and reduced overfitting compared to support vector regression (SVR), partial least squares regression (PLSR), and deep neural network (DNN) models. The models demonstrated the efficacy of hyperspectral data for SOC estimation, suggesting potential for future applications that integrate this data with above-ground biomass to improve SOC mapping across larger scales. This research offers a promising spectral transformation approach for effective carbon management and sustainable agriculture in a changing climate. 

The rapid evolution of the Internet of Things (IoT) has underscored the importance of comprehensive educational strategies to impart IoT concepts and applications to a diverse audience. Given IoT’s pervasive impact, there is hence a pressing need for effective education in this area. Currently, there is a significant gap between existing educational strategies for IoT and the dynamic, engaging approaches needed to captivate a diverse audience, particularly young learners. The challenge lies in developing a methodology that not only educates but also motivates students e.g., from Grade 2 to Grade 12. To address this need, we developed an innovative, activity-based educational framework, integrating interactive and immersive learning methods, aimed at simplifying complex IoT concepts with smart agricultural application in mind for early learners. We outline this novel pedagogical approach, detailing how specific IoT components are taught through targeted activities. The paper should serve as a guide for educators to implement this framework and encourage readers to recognize the importance of adopting new teaching strategies for IoT. Through the imple- mentation of this framework, exemplified in a case study of a plant care game, we have observed an increased engagement and understanding of IoT concepts among our target students. These findings indicate the effectiveness of our approach in real-world educational settings. 

Unmanned aerial vehicle (UAV)-based remote sensing applications in plant phenotyping have received attention in modern plant breeding programs that increasingly have the need to automate time-consuming manual measurements of agronomic traits. This paper focuses on the prediction of sorghum biomass using machine learning algorithms such as Linear Regression, KNeighbors Regressor, and the XGBoost Regressor. Results from a field experiment of 344 sorghum genotypes conducted at the Donald Danforth Plant Science Center (Saint Louis, MO, USA) showed accurate prediction models. The K-Neighbors Regression model performed better than the other two models (R2 = 0.65, RMSE = 4968.60 kg/ha). The developed approach in this study could be used as a decision support tool for sorghum biomass phenotyping in breeding programs. 

We explore the use of deep convolutional neural networks (CNNs) trained on overhead imagery of biomass sorghum to ascertain the relationship between single nucleotide polymorphisms (SNPs), or groups of related SNPs, and the phenotypes they control. We consider both CNNs trained explicitly on the classification task of predicting whether an image shows a plant with a reference or alternate version of various SNPs as well as CNNs trained to create data-driven features based on learning features so that images from the same plot are more similar than images from different plots, and then using the features this network learns for genetic marker classification. We characterize how efficient both approaches are at predicting the presence or absence of a genetic markers, and visualize what parts of the images are most important for those predictions. We find that the data-driven approaches give somewhat higher prediction performance, but have visualizations that are harder to interpret; and we give suggestions of potential future machine learning research and discuss the possibilities of using this approach to uncover unknown genotype × phenotype relationships.