The UF/IFAS Faculty AI Working Group (FAWG) is presenting a series of seminars from faculty across UF/IFAS as well as external speakers to build a robust scholarly community on artificial intelligence (AI) and data science. It will provide a platform for experts to share their latest research advancements in fundamental and applied AI and data science technologies for agriculture, natural resources, and human systems.
Stefano Carpin is a Professor in the Department of Computer Science and Engineering and the Associate Dean for Research and Graduate Programs in the School of Engineering at the University of California, Merced. He received his “Laurea” and Ph.D. degrees in electrical engineering and computer science from the University of Padova (Italy) in 1999 and 2003. From 2003 to 2006, he held faculty positions at Jacobs University Bremen, in Germany. Since 2007, he has been with UC Merced, where he established and leads the UC Merced Robotics laboratory. His research focuses on mobile and cooperative robotics for service tasks and robot algorithms, with an emphasis on applications to precision agriculture and sustainability. Currently, he serves as UC Merced site director for the NSF Engineering Research Center for the Internet Of Things for Precision Agriculture.
Dr. Carpin's research has been supported by the National Science Foundation, DARPA, USDA, the Office of Naval Research, the Army Research Lab, the Department of Commerce (NIST), the Center for Information Technology Research in the Interest of Society (CITRIS), Microsoft Research, and General Motors.
Scaling Data Collection for Precision Agriculture: Challenges and Innovations
The increasing demand for sustainable and resilient agricultural practices has fueled expectations that robotics and AI will play a pivotal role in revolutionizing the industry. While significant advancements have been achieved in both academic research and the corporate sector, numerous critical challenges remain unresolved. In this talk, Stefano Carpin, a professor in computer science and engineering and associate dean for research and graduate programs at the School of Engineering at the University of California, Merced, will present his team’s recent work on addressing one of these challenges -- developing efficient, scalable data collection methods for detecting and managing water stress in California high-value perennial crops. Carpin will discuss the computational and technical hurdles faced in this context and present some solutions that developed and deployed in the field. Finally, he will highlight promising new research directions aimed at reducing barriers to entry for end users, making these technologies more accessible and cost-effective for farmers and agricultural stakeholders.
Dr. Harley is an Associate Professor and a Kent and Linda Fuchs Faculty Fellow in the Department of Electrical and Computer Engineering. He is also the director of the NSF Center for Big Learning. His academic background is in signal processing and artificial intelligence. His research interests integrate physical knowledge with machine learning to advance fundamental sciences and engineering.
Dr. Harley is a recipient of 2021 Achenbach Medal from the International Workshop on Structural Health Monitoring, a 2020 IEEE Ultrasonics, Ferroelectrics, and Frequency Control Society Star Ambassador Award, a 2020 and 2018 Air Force Summer Faculty Fellowship, a 2017 Air Force Young Investigator Award, and a 2014 Carnegie Mellon A. G. Jordan Award (for academic excellence and exceptional service to the community). He has published more than 180 technical journal and conference papers, including four best student papers. He is also an Associate Editor and member of the editorial board for Structural Health Monitoring: An International Journal, a member of the editorial board for Ultrasonics, a member of the IEEE Ultrasonics, Ferroelectrics, and Frequency Control Society Technical Program Committee, a member of the IEEE Signal Processing Society, and a member of the Acoustical Society of America.
Integrating Data-Driven Machine Learning and Knowledge-Driven Analyses for Quantifying Ecosystem Services
Ecosystem services provide the foundation for human well-being and economic development by supporting food production, clean water availability, climate regulation, and other vital functions. Quantifying ecosystem services helps us understand complex relationships between nature and human well-being. This presentation by Joel Harley demonstrates two avenues for creating hybrid machine learning systems. One approach utilizes the Decision Support System for Agrotechnology Transfer (DSSAT) to improve our ability to predict nitrogen levels in soil. The second utilizes ecological relationships between land use and other ecosystem services to improve broader ecosystem service quantification with remote sensing. We discuss how these frameworks can be best utilized as well as the challenges and opportunities associated with them.
Dr. Nikolay Bliznyuk is an Associate Professor of Statistics at the University of Florida, with appointments in the Departments of Agricultural & Biological Engineering, Biostatistics, Statistics and Electrical & Computer Engineering. He earned his doctoral degree in Operations Research & Information Engineering from Cornell University in 2008, concentrating in computational statistics. Prior to joining UF in 2011 as a tenure-track Assistant Professor in the Department of Statistics, he held a postdoctoral researcher appointment in the Department of Biostatistics at the Harvard University and a research assistant professor in the Department of Statistics at Texas A&M University. His research has four tightly intertwined themes: (i) hierarchical Bayesian modeling strategies (ii) spatio-temporal modeling, (iii) methodology and applications of statistical (machine) learning, and (iv) computationally expensive inverse problems (also known as Bayesian calibration of computer models).
Statistical machine learning for improved detection and uncertainty quantification of pathogenic E. Coli in hydroponic irrigation water using impedimetric systems
Reuse of alternative water sources for irrigation is a sustainable approach that can reduce water gaps, while increasing food production. However, when growing fresh produce, this practice increases the risk of bacterial contamination. Thus, rapid and accurate identification of pathogenic organisms such as E. coli is crucial for resource management. Dr. Nikolay Bliznyuk, a University of Florida associate professor of statistics, presented a statistical machine learning framework his team developed to predict and quantify uncertainty about E. coli concentration, which lets scientists quantify uncertainty about the bacterial concentration in a solution sample.
April 22, 2024, 12:00PM to 1:00PM, Department of Animal Sciences, L.E. “Red” Larson Building, Room 201
Dr. Rohan L Fernando
Professor Emeritus - Department of Animal Science, Iowa State University
Dr. Fernando received his Ph.D. degree in Animal Science from the University of Illinois in 1984 and continued there as a faculty member. In 1996, he joined the Department of Animal Science at Iowa State University as a Full Professor. Dr. Fernando is internationally recognized for his contributions to the development of methodology for genetic improvement of livestock, especially in the area of incorporating molecular information into breeding. He has received over 100 invitations to speak and teach on these developments in 21 countries. He retired from Iowa State University in 2022.
How would Fisher use genomic information to determine the genetic contribution to continuous traits?
In 1918, Ronald Fisher developed theory to demonstrate that under the assumption of Mendelian inheritance, one could determine the genetic contribution to continuous traits. In the absence of genomic information, he used pedigree relationships between individuals to determine the covariance between their genotypic values. In most of the current genetic analyses that utilize genomic information, the pedigree relationships are replaced by genomic relationships. In this talk, Rohan Fernando presented an alternative approach to study the genomic contribution to continuous traits that does not require making many of the usual assumptions such as random mating and linkage equilibrium.
April 05, 2024, 12:00PM to 1:00PM, Frazier Rogers R-122
Dr. Ibrahim Volkan Isler
Professor - Computer Science and Engineering, University of Minnesota
Volkan Isler is a professor of computer science and engineering at the University of Minnesota. From 2021 to 2023, he was the head of Samsung’s AI Center in New York City. He is primarily interested in coupling perception and planning in robotics. Volkan has worked on fundamental algorithmic problems in this domain (pursuit-evasion and sensor planning), and developed field systems for environmental sensing, agricultural and home automation. He is the recipient of the NSF CAREER award, UMN McKnight Land-grant Professorship, and the Institute on the Environment Resident Fellowship. He has served as a member of the IEEE RAS Conference Editorial Board, and as an associate editor for both the IEEE Transactions on Robotics, and IEEE Transactions on Automation Science & Engineering. He is co-founder of Farm-Vision Technologies — a UMN start-up based on his lab’s work on yield mapping for specialty farms.
Top-down vs. Bottom-up Intelligence: A Robotics Perspective
Over the past couple of decades, robotics research has mainly focused on transforming classical automation-style robots into intelligent machines that can operate in complex and unstructured environments with minimal human intervention. With recent advances in machine learning, the development of such intelligent robot systems is closer than ever to reality. In this talk, Dr. Volkan Isler gives an overview of robot systems that have been built by his group over the years. He also presents their recent work to combine top-down and bottom-up approaches to intelligence toward building general purpose robots that can operate robustly and safely in everyday environments.
March 22, 2024, 12:00PM to 01:00PM
AI-Powered Phenomics for Plant Breeding
Presented by Xu "Kevin" Wang, University of Florida, Gulf Coast Research and Education Center - Agricultural and Biological Engineering
AI Modeling of Soil Carbon from Local to Global Scales
Presented by Sabine Grunwald, University of Florida - Soil, Water and Ecosystem Sciences
Advancing Plant Breeding through AI-Powered Phenomics
The integration of Artificial Intelligence (AI) in modern agriculture has revolutionized traditional practices by transitioning from labor-intensive operations to automated, user-friendly, and precision-driven approaches. Specifically, in plant breeding, AI-powered phenomics has great potential to effectively increase the genetic gain across breeding cycles. In this presentation, Dr. Kevin Wang, UF/IFAS assistant professor of agricultural and biological engineering delved into two AI-driven phenomics initiatives: the detection of strawberry runners and the estimation of strawberry biomass. These advancements highlight the incredible potential of AI in unlocking new frontiers for crop improvement in plant breeding.
AI Modeling of Soil Carbon from Local to Global Scale
Dr. Sabine Grunwald, a UF/IFAS professor of pedometrics, landscape analysis and geographic information systems in the UF/IFAS soil, water and ecosystem sciences department, talked about the optimization of soil carbon sequestration and reduction of land-based greenhouse gas emissions that are critically important for soil health, food security, carbon quantification and crediting, and resilient and sustainable agro-ecosystems. Whether soil carbon acts as a sink or source for carbon depends on multiple environmental factors that vary across spatial scales (field, regional, and global scale). Data- and sensor-driven approaches using machine learning and deep learning AI (artificial intelligence) have emerged as powerful force to model soil carbon dynamics and patterns in dependence of global climate change, crop types, land use management, and site-specific conditions. In this presentation prominent AI models to predict soil organic carbon and various carbon pools in Florida, the U.S. and at global scale are presented. Critical discussion focused on model accuracy, actual vs. attainable (optimized) soil carbon, and spectral and spatial resolutions of sensors needed to support local land management as well as regionalized carbon accounting, policies, and marketing.
February 23, 2024
Integrative and Explainable AI for Early Detection of Multifactorial Autoimmune Diseases
Presented byRaquel Dias, University of Florida - Microbiology and Cell Science
Dr. Dias explored the transformative impact of artificial intelligence in genomic research, specifically focusing on genotype imputation and mutation screening, the process of identifying genetic mutations or variants in an individual's DNA. Mutation screening can be used to design new genes/proteins with a desired function or to improve stability and function of proteins that already exist. By harnessing advanced machine-learning algorithms and data-driven approaches, scientists have enhanced the accuracy and speed of genotype imputation processes, filling in missing genetic information with unprecedented precision.
November 17, 2023
Applications of Generative Artificial Intelligence Models to Agriculture: Preliminary Studies on Heat Stress Estimation Using Aerial Images
Presented by Henry Medeiros, University of Florida - Agricultural and Biological Engineering
Novel Uses of Generative AI in Agricultural Machine Vision
Presented byArnold Schumann, University of Florida - Soil, Water, and Ecosystem Sciences
Applications of Generative Artificial Intelligence Models to Agriculture: Preliminary Studies on Heat Stress Estimation Using Aerial Images
In this presentation,Henry Medeiros, an associate professor of agricultural and biological engineeringexplored the potential of generative artificial intelligence models to design soft sensors for agricultural applications. In practical terms crop heat stress depends on several environmental factors and has been estimated by using costly thermal aerial imaging. Medeiros presented preliminary results that demonstrate the capability of generative models to estimate heat stress using RGB (color) images.
Novel Uses of Generative AI in Agricultural Machine Vision
Generative AI heralds a novel approach for augmenting sparse datasets with synthetic images for mission-critical deep learning machine vision applications in agriculture. Convolutional artificial neural network (CNN) models used in deep learning image classification require large collections of labeled images for training. In this presentation,Arnold Schumann, a professor of soil, water and ecosystem sciencestalked about how smartphone apps using CNNs to detect exotic plant diseases could help mobilize early responses and quarantines.
October 27, 2023
Convergence of Mechanistic Modeling and Artificial Intelligence in Hydrologic Science and Engineering (and Lessons for Other Fields)
Presented by Rafael Munoz-Carpena, University of Florida - Agricultural and Biological Engineering
Data Ethics in AI Development and Implementation in Agriculture
Presented by Ziwen Yu, University of Florida - Agricultural and Biological Engineering
Convergence of Mechanistic Modeling and Artificial Intelligence in Hydrologic Science and Engineering (and Lessons for Other Fields)
Our water system is complex, and studying it requires analyses of increasingly large data available from conventional and remote sensing and IoT sensor technologies. Some artificial intelligence applications lack the ability to address explicitly important hydrological questions. Rafael Munoz-Carpena, a professor of agricultural and biological engineering, presented four main types of hydrological problems based on their dominant space and time scales. He also identified important opportunities for AI and machine learning in hydrology.
Data Ethics in AI Development and Implementation in Agriculture
Given legal limitations, defining data as a property that can be owned is a tricky and difficult process, especially for farmers. Big companies, with professional legal teams, have already created aggressive contractual terms to protect their rights and obligations. However, farmers, lacking related knowledge and skills, may not realize the problem or, even if they do, may not know how to react. Recent efforts indicate that interpreting data rights from contractual terms is challenging. In his presentation, Dr. Ziwen Yu, a UF/IFAS assistant professor of agricultural and biological engineering, explained key elements regarding the protection of data rights, leveraging data value, and recognizing emerging data-derivative markets.
September 29, 2023
Development of User-Friendly, Open-Source Computer Vision Tools for Precision Livestock Farming
Presented byHaipeng Yu, University of Florida - Animal Science
Automation and Deep Learning to Advance Phenomics and Postharvest Handling
Presented by Charlie Li, University of Florida - Agricultural and Biological Engineering
Development of User-Friendly, Open-Source Computer Vision Tools for Precision Livestock Farming
Haipeng Yu, a UF/IFAS assistant professor of animal sciences and quantitative geneticist, uses computer vision (CV) to help livestock producers. Yu introduced ShinyAnimalCV, an open-source application that provides a user-friendly interface for performing CV tasks. Yu says CV technology optimizes decision-making through timely and individualized animal care. Affordable two- and three-dimensional camera sensors, combined with various algorithms, have provided opportunities to improve livestock production systems. Yu anticipates ShinyAnimalCV can contribute to CV research and teaching in the animal science community.
Automation and Deep Learning to Advance Phenomics and Postharvest Handling
Charlie Li, a UF/IFAS professor of agricultural and biological engineering, talked about sustainably intensifying agricultural production and food supply while preserving the environment. Li will go over multiple research projects that leverage agricultural robotics and deep learning to address challenges spanning the food chain — from breeding to harvest and postharvest handling. He presented a novel modular agricultural robotic system (MARS) that is an autonomous, multi-purpose, and affordable robotic platform for in-field automated phenotyping and precision farming.
August 16, 2023
AI and Machine Learning to Reduce Postharvest Loss
Presented by Tie Liu, University of Florida - Horticultural Sciences
From Soil Mapping to Informed Decision Making for Ecosystem Health: An Envisioned Target and the Role of AI
Presented by Nikos Tziolas, University of Florida - Soil, Water, and Ecosystem Sciences
AI and machine learning to reduce postharvest losses
Tie Liu, an assistant professor of horticultural sciences, says the quality of fresh fruits and vegetables deteriorates before reaching consumers due to biochemical processes and compositional changes. In his presentation Liutalked about how he and his team are addressing food wasteand loss problems.They’re leveraging imaging-based machine learning technologies to understand postharvest deterioration and microbial spoilage of fresh produce to evaluate the quality. Liu proposes a research program to identify proteins and compounds as “freshness indicators” and to help develop easy-to-use tools to accurately estimate the freshness of produce and or contamination of produce. Clickherefor more about Liu’s research.
From soil mapping to informed decision making for ecosystem health: An envisioned target and the role of AI
A new approach to training agricultural robotics through synthetic data and digital twin
Dana Choi,an assistant professor of agricultural and biological engineering at the Gulf Coast Research and Education Center, talked about how AI advancements offer groundbreaking solutions across numerous fields, including agriculture. However, training machine-learning models for agricultural robotics presents significant challenges. She stressed the scarcity of high-quality training data, complex real-world agricultural environments and the time-consuming, costly nature of physical testing. Choi presented her current research in which she leverages synthetic data and digital twins to train machine-learning algorithms. Learn more aboutthis research here.
Geospatial Artificial Intelligence for Ecosystem Service Quantification
At the same seminar, Chang Zhao, anassistant professor of agronomy, presented how her research can advance quantification and mapping ecosystem services, which include ways that human health and well-being are closely tied to the environment. Those include benefits such as food production,carbon storage and sequestration, habitat conservation, and non-material benefits such as recreation and landscape appreciation. She described several research projects she’s leading or working on, including remote sensing and AI-based methods to quantify and map land-cover change, vegetation dynamics and biodiversity. Zhao’s ultimate goal is to help us understand ecosystem services and develop easy-to-use geographic information tools that inform evidence-based decision-making for sustainable land use planning and management.
.png)