Location: Southern Horticultural Research Unit
Project Number: 6062-30500-001-007-S
Project Type: Non-Assistance Cooperative Agreement
Start Date: Apr 7, 2026
End Date: Dec 31, 2028
Objective:
Objective 1. Develop and validate a digital horticulture tool, using UAV-based aerial imagery and artificial intelligence-driven analytics, for automated plant detection and counting for two ornamental crops towards accurate and efficient inventory management in nurseries. Objective 2. Develop and validate a model for ex situ image analysis-based fruit weight estimation for a small fruit crop, a first step towards ex situ and in-field fruit yield estimation.
Objective 3. Develop a comprehensive dataset of aerial imagery for multiple ornamental crops, including the ones from this project, and develop a web portal containing the plant counting tool, where growers can upload images/videos of plants from their nurseries to get the plant counts.
Objective 4: Conduct exploratory research on using UAV-based remote sensing and AI analytics for assessing freeze injury in blueberries
Objective 5: Publish two papers in peer-reviewed journals: one on plant detection and counting and the other a data resource paper. In addition, present and publish a paper in a top national or international precision/digital agriculture conference.
Approach:
We will employ NDAA approved / Blue UAS certified unmanned aerial vehicles (UAVs) for collecting aerial images of ornamental plants in nurseries located in Mississippi, Alabama, and Louisiana. The images will be collected under natural daylight conditions with UAVs flying at a low altitude (~30-40 ft) using RGB and multispectral cameras. Datasets of digitally processed images will be used for developing computational models using different AI analytics to realize higher accuracy of plant counting in each of the target ornamental crops. The accuracy and utility of the tool will be validated by working with nursery owners. We will create a comprehensive dataset of aerial imagery from this as well as previous USDA- cooperator projects and will be made available to public, along with the counting tool, through a web portal. Nursery owners and growers can use the portal to upload images/videos of their ornamental crops and instantly get the plant counts. We will also gather aerial images of blueberries and build AI-based models to assess the extent of freeze injury. This complements ongoing work at ARS Poplarville on assessing the efficacy of different nanomaterials to offer protection against freeze injury. The goal is to routinely use aerial imaging based digital approach for assessing freeze injury so that growers can take efficient management decisions for their blueberry farms and researchers can use it in germplasm and variety/hybrid development programs. We will publish the tool development work and the data resource papers in high-impact peer-reviewed journals.