Agronomy Center For Research And Education Research Articles

Structural Component Phenotypic Traits from Individual Maize Skeletonization by UAS-Based Structure-from-Motion Photogrammetry

The bottleneck in plant breeding programs is to have cost-effective high-throughput phenotyping methodologies to efficiently describe the new lines and hybrids developed. In this paper, we propose a fully automatic approach to overcome not only the individual maize extraction but also the trait quantification challenge of structural components from unmanned aerial system (UAS) imagery. The experimental setup was carried out at the Indiana Corn and Soybean Innovation Center at the Agronomy Center for Research and Education (ACRE) in West Lafayette (IN, USA). On 27 July and 3 August 2021, two flights were performed over maize trials using a custom-designed UAS platform with a Sony Alpha ILCE-7R photogrammetric sensor onboard. RGB images were processed using a standard photogrammetric pipeline based on structure from motion (SfM) to obtain a final scaled 3D point cloud of the study field. Individual plants were extracted by, first, semantically segmenting the point cloud into ground and maize using 3D deep learning. Secondly, we employed a connected component algorithm to the maize end-members. Finally, once individual plants were accurately extracted, we robustly applied a Laplacian-based contraction skeleton algorithm to compute several structural component traits from each plant. The results from phenotypic traits such as height and number of leaves show a determination coefficient (R2) with on-field and digital measurements, respectively, better than 90%. Our test trial reveals the viability of extracting several phenotypic traits of individual maize using a skeletonization approach on the basis of a UAS imagery-based point cloud. As a limitation of the methodology proposed, we highlight that the lack of plant occlusions in the UAS images obtains a more complete point cloud of the plant, giving more accuracy in the extracted traits.

Agricultural Aerosols: The Impact of Farming Activity on Ice Nucleating Particles

Farming activities cause particles such as soil dust and plant material to be emitted into the air. Some of these aerosols can become ice nucleating particles (INPs), serving as seeds for ice and mixed-phase clouds. While there have been ground-based studies of these particles in the western Great Plains and a single air-based study in Indiana, there is a distinct lack of ground-based studies in the Midwest. In Indiana, over two-thirds of the state is farmland, with over 75% of land in Tippecanoe County used for agriculture. Despite farming being such an essential part of life in Indiana, the connection between agricultural activities and INP concentrations in the area has not been explored. Using field observations taken at the Purdue Agronomy Center for Research and Education (ACRE), we hope to study the impact of harvesting on INP concentrations in the midwestern United States. The field experiment took place from May to December 2021 at the ACRE site, but this study focuses on three days during the harvesting period. Data was collected via two instruments: the SPectrometer for Ice Nuclei (SPIN) and the Cloud Condensation Nuclei Counter (CCNC). It appears there is an increase in INP concentrations on days when harvesting occurs, most likely due to an increase in organic and biological particles. It is hoped that the data from this project will provide further insight into the composition and number concentrations of INPs from harvesting through ground-based field observations, as well as insight into INP concentration in the rural Midwest and its climatic impacts.

GeoDLS: A Deep Learning-Based Corn Disease Tracking and Location System Using RTK Geolocated UAS Imagery

Deep learning-based solutions for precision agriculture have recently achieved promising results. Deep learning has been used to identify crop diseases at the initial stages of disease development in an effort to create effective disease management systems. However, the use of deep learning and unmanned aerial system (UAS) imagery to track the spread of diseases, identify diseased regions within cornfields, and notify users with actionable information remains a research gap. Therefore, in this study, high-resolution, UAS-acquired, real-time kinematic (RTK) geotagged, RGB imagery at an altitude of 12 m above ground level (AGL) was used to develop the Geo Disease Location System (GeoDLS), a deep learning-based system for tracking diseased regions in corn fields. UAS images (resolution 8192 × 5460 pixels) were acquired in cornfields located at Purdue University’s Agronomy Center for Research and Education (ACRE), using a DJI Matrice 300 RTK UAS mounted with a 45-megapixel DJI Zenmuse P1 camera during corn stages V14 to R4. A dataset of 5076 images was created by splitting the UAS-acquired images using tile and simple linear iterative clustering (SLIC) segmentation. For tile segmentation, the images were split into tiles of sizes 250 × 250 pixels, 500 × 500 pixels, and 1000 × 1000 pixels, resulting in 1804, 1112, and 570 image tiles, respectively. For SLIC segmentation, 865 and 725 superpixel images were obtained using compactness (m) values of 5 and 10, respectively. Five deep neural network architectures, VGG16, ResNet50, InceptionV3, DenseNet169, and Xception, were trained to identify diseased, healthy, and background regions in corn fields. DenseNet169 identified diseased, healthy, and background regions with the highest testing accuracy of 100.00% when trained on images of tile size 1000 × 1000 pixels. Using a sliding window approach, the trained DenseNet169 model was then used to calculate the percentage of diseased regions present within each UAS image. Finally, the RTK geolocation information for each image was used to update users with the location of diseased regions with an accuracy of within 2 cm through a web application, a smartphone application, and email notifications. The GeoDLS could be a potential tool for an automated disease management system to track the spread of crop diseases, identify diseased regions, and provide actionable information to the users.

Comparison of Deep Learning Models for Corn Disease Region Location, Identification of Disease Type, and Severity Estimation Using Images Acquired From UAS-Mounted and Handheld Sensors

Highlights An approach using deep learning was proposed for identifying diseased regions in UAS imagery of corn fields with 97.23% testing accuracy using the VGG16 model. Disease types were identified within the diseased regions with a testing accuracy of 98.85% using the VGG16 model. On the diseased leaves, severity was estimated with a testing accuracy of 94.20% using the VGG16 model. Deep Learning models have the potential to bring efficiency and accuracy to field scouting. Abstract. Accurately locating diseased regions, identifying disease types, and estimating disease severity in corn fields are all connected steps for developing an effective disease management system. Traditional disease management that relied on a manual scouting approach was inefficient. Therefore, the research community is working on developing advanced disease management systems using deep learning. However, most of the past studies used public datasets consisting of images with uniform backgrounds acquired under lab conditions to train deep learning models, thus, limiting their use under field conditions. In addition, limited studies have been conducted for in-field corn disease analysis using Unmanned Aerial System (UAS) imagery. Therefore, UAS and handheld imagery sensors were used in this study to acquire corn disease images from fields located at Purdue University’s Agronomy Center for Research and Education (ACRE) in the summer of 2020. A total of 55 UAS flights were conducted over three different corn fields from June 20 through September 29, resulting in a collection of approximately 59,000 images. A novel three-stage approach was proposed by independently training a total of nine image classification models using three neural network architectures, namely: VGG16, ResNet50, and InceptionV3, for locating diseased regions, identifying disease types, and estimating disease severity under field conditions. Diseased regions were first identified accurately in UAS-acquired corn field imagery by a sliding window and deep learning-based image classification, with testing accuracies of up to 97.23%. Diseased region identification was followed by accurately identifying three common corn diseases, namely Northern Leaf Blight (NLB), Gray Leaf Spot (GLS), and Northern Leaf Spot (NLS), within the diseased regions with testing accuracies of up to 98.85%. Finally, the severity of the NLS disease on leaves was estimated with a testing accuracy of up to 94.20%. The VGG16 model achieved the highest testing accuracies for identifying diseased regions in corn fields, identifying corn disease types, and estimating NLS's severity. This study presents promising results for three main elements of a disease management system and could advance traditional scouting by integrating deep learning with UAS imagery. Keywords: Corn Diseases, Datasets, Deep Learning, Disease Identification, Disease Region Location, Image Classification, Severity Estimation, UAS Imagery.

Multi-Temporal Predictive Modelling of Sorghum Biomass Using UAV-Based Hyperspectral and LiDAR Data

High-throughput phenotyping using high spatial, spectral, and temporal resolution remote sensing (RS) data has become a critical part of the plant breeding chain focused on reducing the time and cost of the selection process for the “best” genotypes with respect to the trait(s) of interest. In this paper, the potential of accurate and reliable sorghum biomass prediction using visible and near infrared (VNIR) and short-wave infrared (SWIR) hyperspectral data as well as light detection and ranging (LiDAR) data acquired by sensors mounted on UAV platforms is investigated. Predictive models are developed using classical regression-based machine learning methods for nine experiments conducted during the 2017 and 2018 growing seasons at the Agronomy Center for Research and Education (ACRE) at Purdue University, Indiana, USA. The impact of the regression method, data source, timing of RS and field-based biomass reference data acquisition, and the number of samples on the prediction results are investigated. R2 values for end-of-season biomass ranged from 0.64 to 0.89 for different experiments when features from all the data sources were included. Geometry-based features derived from the LiDAR point cloud to characterize plant structure and chemistry-based features extracted from hyperspectral data provided the most accurate predictions. Evaluation of the impact of the time of data acquisition during the growing season on the prediction results indicated that although the most accurate and reliable predictions of final biomass were achieved using remotely sensed data from mid-season to end-of-season, predictions in mid-season provided adequate results to differentiate between promising varieties for selection. The analysis of variance (ANOVA) of the accuracies of the predictive models showed that both the data source and regression method are important factors for a reliable prediction; however, the data source was more important with 69% significance, versus 28% significance for the regression method.

Yield prediction by machine learning from UAS-based multi-sensor data fusion in soybean

BackgroundNowadays, automated phenotyping of plants is essential for precise and cost-effective improvement in the efficiency of crop genetics. In recent years, machine learning (ML) techniques have shown great success in the classification and modelling of crop parameters. In this research, we consider the capability of ML to perform grain yield prediction in soybeans by combining data from different optical sensors via RF (Random Forest) and XGBoost (eXtreme Gradient Boosting). During the 2018 growing season, a panel of 382 soybean recombinant inbred lines were evaluated in a yield trial at the Agronomy Center for Research and Education (ACRE) in West Lafayette (Indiana, USA). Images were acquired by the Parrot Sequoia Multispectral Sensor and the S.O.D.A. compact digital camera on board a senseFly eBee UAS (Unnamed Aircraft System) solution at R4 and early R5 growth stages. Next, a standard photogrammetric pipeline was carried out by SfM (Structure from Motion). Multispectral imagery serves to analyse the spectral response of the soybean end-member in 2D. In addition, RGB images were used to reconstruct the study area in 3D, evaluating the physiological growth dynamics per plot via height variations and crop volume estimations. As ground truth, destructive grain yield measurements were taken at the end of the growing season.ResultsAlgorithms and feature extraction techniques were combined to develop a regression model to predict final yield from imagery, achieving an accuracy of over 90.72% by RF and 91.36% by XGBoost.ConclusionsResults provide practical information for the selection of phenotypes for breeding coming from UAS data as a decision support tool, affording constant operational improvement and proactive management for high spatial precision.

HIGH THROUGHPUT PHENOTYPING OF PHYSIOLOGICAL GROWTH DYNAMICS FROM UAS-BASED 3D MODELING IN SOYBEAN

Abstract. Nowadays, an essential tool to improve the efficiency of crop genetics is automated, precise and cost-effective phenotyping of the plants. The aim of this study is to generate a methodology for high throughput phenotyping the physiological growth dynamics of soybeans by UAS-based 3D modelling. During the 2018 growing season, a soybean experiment was performed at the Agronomy Center for Research and Education (ACRE) in West-Lafayette (Indiana, USA). Periodic images were acquired by G9X Canon compact digital camera on board senseFly eBee. The study area is reconstructed in 3D by Image-based modelling. Algorithms and techniques were combined to analyse growth dynamics of the crop via height variations and to quantify biomass. Results provide practical information for the selection of phenotypes for breeding.

Automated Identification of Tile Lines from Remotely Sensed Data

Although subsurface drainage provides many agronomic and environmental benefits, extensive subsurface drainage systems have important implications for surface water quality and hydrology. Due to limited information on subsurface drainage extent, it is difficult to understand the hydrology of intensively tile-drained watersheds. In order to address this problem, a methodology was developed to use image processing techniques for automated detection of tile drains from multiple dates of aerial photography at the Agronomy Center for Research and Education (ACRE), West Lafayette, Indiana. A stepwise approach was adopted to first identify potential tile-drained fields from the GIS-based analysis of land use class, soil drainage class, and surface slope using decision tree classification. Based on preliminary classification of potential tile-drained area from the decision tree classifier, a combination of image processing techniques such as directional edge enhancement filtering, density slice classification, Hough transformation, and automatic vectorization were used to identify individual tile lines from images of 1976, 1998, and 2002. Accuracy assessment of the predicted tile line maps (Hough transformed and untransformed) was accomplished by comparing the locations of predicted tile lines with the known tile lines mapped through manual digitization from historic design diagrams using both a confusion matrix approach and drainage density. Forty-eight percent of tile lines were correctly predicted for the Hough transformed map and 58% for the untransformed map based on the producer accuracy. Similarly, 73% of non-tile area was correctly predicted for Hough transformed and 68% for untransformed lines. Based on drainage density calculation, 60% of tile lines were predicted from the aerial image of 1976 and 50% from the aerial image of 2002 for both techniques, while 72% of tile lines were predicted from the aerial image of 1998 for untransformed and 50% for Hough transformed lines. The Hough transformation provided the best results in producing a map without discontinuity between lines. The overall performance of the image processing techniques used in this study shows that these techniques can be successfully applied to identify tile lines from aerial photographs over a large area.