2024 Seed Grant Recipients

SSIF: A Deep Generative Model for Remote Sensing Image Spatial-Spectral Super-Resolution for Precision Agriculture

PI: Gengchen Mai, Assistant Professor, Department of Geography, Athens Campus

Co-PIs: Tianming Liu, School of Computing; Lilong Chai, Department of Poultry Science; Stefano Ermon, Department of Computer Science, Stanford University; Ni Lao, Google LLC; Hongxu Ma, Mineral Earth Sciences LLC, Alphabet; Jinmeng Rao, Mineral Earth Sciences LLC, Alphabet; and Jiaming Song, Luma AI

Project Description: In order to develop a cost-efficient long-term large-scale remote sensing-based surveillance for precision agriculture, we propose a deep generative models for remote sensing spatial-spectral super-resolution framework called Spatial-Spectral Implicit Function (SSIF). The advantage of SSIF is that SSIF can jointly handle remote sensing images with various spatial resolutions and spectral resolutions which makes it a sensor-agnostic model that can be trained on RS image super-resolution datasets based on multiple satellite sensors. We will evaluate our SSIF framework on two agriculture tasks: 1) fine-grained crop classification and crop yield prediction, 2) broiler chicken house counting.

Harnessing Deep Learning Algorithms for Ultrasound Image Analysis to Support Reproductive Management Decisions in Cattle

PI: Todd Callaway, Associate Professor, Department of Animal & Dairy Science, Athens Campus

Co-PIs: Pedro Fontes, Department of Animal & Dairy Science and Anderson Alves, Department of Animal & Dairy Science

Project Description: The constant advances in bovine ultrasound have contributed to the spread of this technology as a common diagnostic method to facilitate assisted reproduction in cattle. While significant progress has been made, applying artificial intelligence techniques to bovine reproductive ultrasonography could facilitate image interpretation, decrease misdiagnosis, and optimize workflow on the farm. In this project, we will develop computer vision systems using deep learning algorithms to process and classify ultrasound images, aiming to generate on a large scale, automated image-driven indicators that can be used to support reproductive management decisions in cattle.

EggPicker: A mobile robotic system for identifying, localizing, and collecting floor eggs in cage-free hen housing environments

PI: Guoming Li, Assistant Professor, Department of Poultry Science, Athens Campus

Co-PIs: Ramviyas Nattanmai Parasuraman,Computing; and Ramana Pidaparti, Environmental, Civil, Agricultural, and Mechanical Engineering

Project Description: Mobile ground robotic system for collecting floor eggs help to reduce labor needs for manual floor egg collection and increase production efficiency and biosafety in cage-free hen housing systems. The objective of the proposal is to develop a mobile robotic system with a legged robotic platform and robot arm for identifying, localizing, and collecting floor eggs in cage-free hen housing environments.

Enhancing Efficacy of Ornamental Crop Production: Application of Advanced Technologies for Automating Nursery Production Operations to Improve Labor Use Efficiency

PI: Md Sultan Mahmud, Assistant Professor, Department of Plant Pathology, Athens Campus

Co-PIs: Guoyu Lu, College of Electrical and Computer Engineering; Jean Williams-Woodard, Department of Plant Pathology; and Ping Yu, Department of Horticulture

Project Description: The project aims are to revolutionize ornamental crop production by implementing advanced technologies to automate nursery production operations. By leveraging automation, the project seeks to enhance labor efficiency, addressing challenges in the labor-intensive nature of the industry. The innovation approach is expected to optimize resource utilization, improve overall productivity, and contribute to the sustainability of ornamental crop cultivation. If successful, this initiative could set a precedent for incorporating cutting-edge technologies in agriculture to meet growing demand while minimizing environmental impact.

PRECISIONDETECT (PD): AI-Enhanced Hyperspectral Imaging for Early Chicken Egg Fertility Detection

PI: Christopher Kucha, Assistant Professor, Department of Food Science & Technology, Athens Campus

Co-PIs: Prashant Doshi, School of Computing; Anand Mohan, Department of Food Science & Technology; Guoming Li, Department of Poultry Science; and Fanbin Kong, Department of Food Science & Technology

Project Description: The disposal of unhatched chicken eggs causes significant economic loss to poultry farmers and has a severe environmental impact. To address this issue, PRECISIONDETECT aims to develop a non-destructive, rapid, and high-throughput egg fertility detection system using a hyperspectral imaging camera, artificial intelligence, and robotics. The project seeks to overcome the challenges associated with the current candling method of egg fertility detection, which is labor-intensive, subjective, and lacks precision. By improving breeding efficiency and reducing resource wastage, this project will benefit the poultry industry economically and environmentally while, overall, reducing the consequences of food insecurity.

Revolutionizing Agriculture with Data-Driven Solutions: Exploring Microbiome Function and Modulation in Animals through Machine Learning

PI: Aditya Mishra, Assistant Professor, Department of Statistics, Athens Campus

Co-PIs: Abhinav Mishra, Department of Food Science & Technology; Woo Kim, Department of Poultry Science; Michael Rothrock, Egg & Poultry Production Safety Research Unit – USDA

Project Description: This study aims to use innovative machine learning methods to optimize farm practices for improved farm animal microbiomes and enhanced health outcomes. The initial phase focuses on understanding how farming practices and environmental factors influence microbiota, utilizing advanced machine learning to analyze diverse datasets. The collaborative effort, involving experts from various disciplines, contributes to the overarching goal of Integrative Precision Agriculture to refine decision-making processes and optimize field operations in Georgia’s primary industry.

Prediction model for breeding peanuts with desired roasted flavor

PI: Joonhyuk Suh, Assistant Professor, Department of Food Science & Technology, Athens Campus

Co-PIs: Nino Brown, Department of Crop & Soil Sciences; Koushik Adhikari, Department of Food Science & Technology; and Abhinav Mishra, Department of Food Science & Technology

Project Description: Researchers will address a current challenge in the flavor breeding of crops, demanding multiple steps of sensory evaluation and chemical analysis. The team will develop a rapid and reliable prediction model (prototype) based solely on chemical information for the evaluation of roasted flavor in peanuts. Collected information from this seed grant will be used as preliminary data for federal grant applications (e.g., USDA-NIFA) in this area. The developed model will be the basis of big data-driven future models for selecting peanut breeding lines with desired roasted flavor in a short time with less labor and cost, which contribute to the “precision breeding” of peanuts.