No description available.

No description available.

This research uncovers a novel pathway in precision agriculture, emphasizing the utilization of advanced supervised and self-supervised deep learning approaches for an innovative solution to weed detection and crop yield prediction. The study focuses on key weed species: Italian ryegrass in wheat, Palmer amaranth, and common ragweed in soybean, which are troublesome weeds in the United States. One of the most innovative components of this research is the debut of a self-supervised learning approach specifically tailored for soybean yield prediction using only unlabeled RGB images. This novel strategy presents a departure from traditional yield prediction methods that consider multiple variables, thus offering a more streamlined and efficient methodology that presents a significant contribution to the field. To address the monitoring of Italian ryegrass in wheat cultivation, a bespoke Convolutional Neural Network (CNN) model was developed. It demonstrated impressive precision and recall rates of 100% and 97.5% respectively, in accurately classifying Italian ryegrass in the wheat. Among three hyperparameter tuning methods, Bayesian optimization emerges as the most efficient, delivering optimal results in just 10 iterations, contrasting with 723 and 304 iterations required for grid search and random search respectively. Further, this study examines the performance of various classification and object detection algorithms on Unmanned Aerial Systems (UAS)-acquired images at different growth stages of soybean and Palmer amaranth. Both the Vision Transformer and EfficientNetB0 models display promising test accuracies of 97.69% and 93.26% respectively. However, considering a balance between speed and accuracy, YOLOv6s emerged as the most suitable object detection model for real-time deployment, achieving an 82.6% mean average precision (mAP) at an average inference speed of 8.28 milliseconds. Furthermore, a self-supervised contrastive learning approach was introduced for automating the labeling of Palmer amaranth and soybean. This method achieved a notable 98.5% test accuracy, indicating the potential for cost-efficient data acquisition and labeling to advance precision agriculture research. A separate study was conducted to detect common ragweed in soybean crops and the prediction of soybean yield impacted by varying weed densities. The Vision Transformer and MLP-Mixer models achieve test accuracies of 97.95% and 96.92% for weed detection, with YOLOv6 outperforming YOLOv5, attaining an mAP of 81.5% at an average inference speed of 7.05 milliseconds. Self-supervised learning-based yield prediction models reach a coefficient of determination of up to 0.80 and a correlation coefficient of 0.88 between predicted and actual yield. In conclusion, this research elucidates the transformative potential of self-supervised and supervised deep learning techniques in revolutionizing weed detection and crop yield prediction practices. Its findings significantly contribute to precision agriculture, paving the way for efficient and cost-effective site-specific weed management strategies. This, in turn, promotes reduced environmental impact and enhances the economic sustainability of farming operations. / Master of Science in Life Sciences / This novel research provides a fresh approach to overcoming some of the biggest challenges in modern agriculture by leveraging the power of advanced artificial intelligence (AI) techniques. The study targets key disruptive weed species, such as, Italian ryegrass in wheat, Palmer amaranth, and common ragweed in soybean, all of which have the potential to significantly reduce crop yields. The studies were first conducted to detect Italian ryegrass in wheat crops, utilizing RGB images. A model is built using a complex AI system called a Convolutional Neural Network (CNN) to detect this weed with remarkable accuracy. The study then delves into the use of drones to take pictures of different growth stages of soybean and Palmer amaranth plants. These images were then analyzed by various AI models to assess their ability to accurately identify the plants. The results show some promising findings, with one model being quick and accurate enough to be potentially used in real-time applications. The most important part of this research is the application of self-supervised learning, which learns to label Palmer amaranth and soybean plants on its own. This novel method achieved impressive test accuracy, suggesting a future where data collection and labeling could be done more cost-effectively. In another related study, we detected common ragweed in soybean crops and predicted soybean yield based on various weed densities. AI models once again performed well for weed detection and yield prediction tasks, with self-supervised models showcasing high agreement between predicted and actual yields. In conclusion, this research showcases the exciting potential of self-teaching and supervised AI in transforming the way we detect weeds and predict crop yields. These findings could potentially lead to more efficient and cost-effective ways of managing weeds at specific sites. This could have a positive impact on the environment and improve the economic sustainability of farming operations, paving the way for a greener future.

The population of the world is increasing exponentially each year with a large population base. Agricultural fields are facing the pressure of dealing with food insufficiency, whereas the challenges of limited resources of arable land and fresh water on the earth should be taken into account at the same time. Smart farming was born at the right time to cope with the problem and has become one of the most powerful approaches to reducing the ecological footprint of farming and improving agricultural yield. The four most important variables that impact crop yield are soil productivity, the accessibility of water, climate, and pests or diseases. This thesis emphasizes the application of chemical fertilizers to corn and disregards the impact of water, pests, and disease for the moment. In this study, three scenarios are explored deeper one by one. The only factor that varies among the three scenarios is the nitrogen amount available to the plant. Fertilizers have outstanding performance in improving the yield and quality of plants in agricultural fields, and this is the emphasis of this thesis. Compared with the fertilizer properties and characteristics of frequently used commercial fertilizers, polymer-coated urea was selected as the fertilizer in this study because the feature of nitrogen can be released into the soil slowly and in a controlled manner. Scenario 1 created an ideal condition where unlimited nitrogen was provided to the corn. Scenario 2 assumed that a fixed amount of polymer-coated urea was applied at the beginning of the sowing season only. Scenario 3 figured out an optimal yield by separating the fertilizer application at the beginning and in the middle of the growing days with the same amounts of fertilizer used in Scenario 2. The model was performed based on historical data from Oklahoma and Ottawa using IoT sensors. The simulation model generated with Python figured out that approximately the end of June to the start of July is the best time to apply the remaining fertilizer, assuming that the sowing stage starts on May 1. The percentage of polymer-coated urea applied initially was found to usually be around 10% in the tested regions. The model was used to predict the yield in Ottawa using from 40.94 g/(m^2) in Scenario 2 to 55.43 g/(m^2) in Scenario 3, achieving an outstanding increasing rate of 35.38%.

By 2050, food consumption and agricultural water use will increase as a result of a global population that is projected to reach 9 billion people. To address this food and water security challenge, there has been increased attention towards the concept of sustainable agriculture, which has a broad aim of securing food and water resources while preserving the environment for future generations. An element of this is the use of precision agriculture, which is designed to provide the right inputs, at the right time and in the right place. In order to optimize nutrient application, water intakes, and the profitability of agricultural areas, it is necessary to improve our understating and predictability of agricultural systems at high spatio-temporal scales. The underlying goal of the research presented herein is to advance the monitoring of croplands and crop yield through high-resolution satellite data. In addressing this, we explore the utility of daily CubeSat imagery to produce the highest spatial resolution (3 m) estimates of leaf area index and crop water use ever retrieved from space, providing an enhanced capacity to provide new insights into precision agriculture. The novel insights on crop health and conditions derived from CubeSat data are combined with the predictive ability of crop models, with the aim of improving crop yield predictions. To explore the latter, a sensitivity analysis-linked Bayesian inference framework was developed, offering a tool for calibrating crop models while simultaneously quantifying the uncertainty in input parameters. The effect of integrating higher spatio-temporal resolution data in crop models was tested by developing an approach that assimilates CubeSat imagery into a crop model for early season yield prediction at the within-field scale. In addition to satellite data, the utility of even higher spatial resolution products from unmanned aerial vehicles was also examined in the last section of the thesis, where future research avenues are outlined. Here, an assessment of crop height is presented, which is linked to field biomass through the use of structure from motion techniques. These results offer further insights into small-scale field variabilities from an on-demand basis, and represent the cutting-edge of precision agricultural advances.

The catalysis reaction performance, such as yield, is very crucial in organic chemistry. And predicting the reaction yield is still very challenging. In this thesis, machine learning is used to predict the reaction yield in a C–N cross-coupling approach. The reaction data are from the high-throughput experimental data with four variables: reactants, Pd catalysts, additives, and bases. Each reaction data will give the corresponding yield. The data are from the literature, which has been uploaded. The total data number used in machine learning is 7910. The method mainly consists of four steps. First, load the csv data and import modules. Second, encode data with molecular fingerprint or one-hot encoding. The data will be normalized if there is need. Third, split the dataset into train and test set with the size ratio of 7/3 or 8/2. Fourth, use six machine learning models to learn the data and evaluate their performance. Then, compare the prediction yield of the test set. The accuracy in prediction (RMSE value and R-squared) and running time will be considered for evaluation. By comparing the RMSE and R-squared values of different models, we can decide which one has better performance and better fitting results. Improved reaction performance, or high-performance catalysts and their characteristics may be obtained.

Doctor of Philosophy / Department of Geography / Douglas G. Goodin / Jesse A. Poland / Kevin Price / Wheat breeders are in a race for genetic gain to secure the future nutritional needs of a growing population. Multiple barriers exist in the acceleration of crop improvement. Emerging technologies are reducing these obstacles. Advances in genotyping technologies have significantly decreased the cost of characterizing the genetic make-up of candidate breeding lines. However, this is just part of the equation. Field-based phenotyping informs a breeder’s decision as to which lines move forward in the breeding cycle. This has long been the most expensive and time-consuming, though most critical, aspect of breeding. The grand challenge remains in connecting genetic variants to observed phenotypes followed by predicting phenotypes based on the genetic composition of lines or cultivars. In this context, the current study was undertaken to investigate the utility of UAS in assessment field trials in wheat breeding programs. The major objective was to integrate remotely sensed data with geospatial analysis for high throughput phenotyping of large wheat breeding nurseries. The initial step was to develop and validate a semi-automated high-throughput phenotyping pipeline using a low-cost UAS and NIR camera, image processing, and radiometric calibration to build orthomosaic imagery and 3D models. The relationship between plot-level data (vegetation indices and height) extracted from UAS imagery and manual measurements were examined and found to have a high correlation. Data derived from UAS imagery performed as well as manual measurements while exponentially increasing the amount of data available. The high-resolution, high-temporal HTP data extracted from this pipeline offered the opportunity to develop a within season grain yield prediction model. Due to the variety in genotypes and environmental conditions, breeding trials are inherently spatial in nature and vary non-randomly across the field. This makes geographically weighted regression models a good choice as a geospatial prediction model. Finally, with the addition of georeferenced and spatial data integral in HTP and imagery, we were able to reduce the environmental effect from the data and increase the accuracy of UAS plot-level data. The models developed through this research, when combined with genotyping technologies, increase the volume, accuracy, and reliability of phenotypic data to better inform breeder selections. This increased accuracy with evaluating and predicting grain yield will help breeders to rapidly identify and advance the most promising candidate wheat varieties.

In order to reduce the vulnerability of wheat production to drought, a calibrated and validated CERES Wheat crop simulation model was used to predict wheat yields on major soil textural groups using historical weather data at Swift Current, Saskatoon and Melfort. Yields were predicted using a run-out technique which involved the use of actual weather data to the prediction date and historical weather data from 1960 to 1990 for the remainder of the growing season. Yield predictions were made at five Julian dates during the crop calendar and these dates coincided with crop emergence, terminal spikelet initiation, end of the vegetative growth, heading and start of grain filling. Three sample years were used as case studies to test the applicability of the run-out method in making yield predictions. Sample base years were those with the lowest, medium and highest yields between 1960 and 1990 and these were selected from ranked yield values using quartiles. Test years were termed base years and weather files that were joined with the test years were run-out years. Each base year had 30 run-out years (1960-1990) and the mean of each run-out year was compared with the observed yield at the end of the season. Run-out yields for each base year were summarised as simple probability distributions so that yields exceeding certain values could be selected. Run-out yields at five prediction dates were found to be in close agreement with observed yields at the end of the growing season. To account for the variability in yields that can be found between places within the same climatic zone, simulated yields were re-classified by soil type and water stress level. These modifiers (soil type and water stress level) showed that chances of getting high yields diminish from Melfort to Swift Current at all prediction points due to the high variability of yield factors. Yield predictions that were made as above suggested that if historical weather records are combined with available weather data during the growing season, a good indication of yields can be obtained ahead of the harvest time and this could allow producers and those in the agri-business to decide on alternative actions of minimizing losses when prospects of getting a good yield are poor.

<p>Improved phenotyping technologies and data analytic strategies have the potential to reduce the phenotyping bottleneck in breeding programs, increase the number of genotypes that can be evaluated, and improve genetic gain of maize. Ear photometry and remote sensing were evaluated in this dissertation for their integration into breeding programs to understand the development of grain yield in diverse germplasm and to better predict yield performance. Ear photometry was used in Chapter 2 to characterize the testcross performance of temperate and tropical inbred lines. The effect of heterosis among the temperate heterotic groups was more noticeable in the ear-related characteristics rather than kernel-size characteristics. Yield components were generally more heritable than grain yield per ear, so they were explored for their use in multi-trait genomic prediction for grain yield on a plot or ear basis in Chapter 3. Multi-trait genomic prediction of grain yield was improved where ear characteristics were known in the testing set compared with single-trait genomic prediction.  Additionally, single-trait genomic prediction was more accurate in the temperate germplasm compared to the tropical germplasm. Thus, ear photometry is an efficient method to quickly assess yield components in maize and improve yield prediction in certain circumstances. In Chapter 4, the effect of row selection, plot size, and plot trimming on remote sensing trait repeatability and prediction accuracy of biomass yield in sorghum or grain yield in maize was evaluated. Decreased plot size and configuration has been suggested to increase the number of genotypes that can be evaluated per unit area. In this study, larger plot sizes were favorable for increasing repeatability and excluding outer rows improved predictive modelling. Plot trimming was never shown to be significantly different from non-trimmed plots in this study. Genomic prediction is another way to minimize experimental size and phenotypic data collection and was evaluated in Chapter 5. A reaction norm was used to model the trajectory of hybrid yield performance across a gradient of 86 environments. The heritability and prediction accuracy of grain yield were both improved in the higher-yielding environments compared to the lower-yielding environments. Single nucleotide polymorphisms with the highest magnitude of effects were selected in each environment. Twenty-one SNPs were selected indicating many SNPs were selected in multiple environments. Candidate genes in linkage disequilibrium with many of these SNPs were previously reported as stress adaptions. Genomic prediction and remote sensing were integrated for prediction of grain yield in Chapter 6. Heritability of remote sensing traits generally improved throughout the growing season. Prediction accuracy of BLUPs were improved through an integrated phenomic and genomic prediction model for all scenarios tested. In summary, ear photometry and remote sensing are technologies to evaluate large populations for unique plant trait characteristics that can be used in combination with genomic prediction to improve understanding of grain yield development and grain yield prediction.</p>

Traditional image processing solutions have been applied in the past to detect and count strawberries. These methods typically involve feature extraction followed by object detection using one or more features. Some object detection problems can be ambiguous as to what features are relevant and the solutions to many problems are only fully realized when the modern approach has been applied and tested, such as deep learning. In this work, we investigate the use of R-CNN for green strawberry detection. The object detection involves finding regions of interest (ROIs) in field images using the selective segmentation algorithm and inputting these regions into a pre-trained deep neural network (DNN) model. The convolutional neural networks VGG, MobileNet and ResNet were implemented to detect subtle differences between green strawberries and various background elements. Downscaling factors, intersection over union (IOU) thresholds and non-maxima suppression (NMS) values can be tweaked to increase recall and reduce false positives while data augmentation and negative hardminging can be used to increase the amount of input data. The state of the art model is sufficient in locating the green strawberries with an overall model accuracy of 74%. The R-CNN model can then be used for crop yield prediction to forecast the actual red strawberry count one week in advance with a 90% accuracy.