University of Delaware FY25 Final Report
Project Title: Development and Evaluation of a Computer Vision–Based
Monitoring Tool for Cucumber Beetles
Executive Summary
This project evaluated the feasibility of using computer vision–based artificial
intelligence (AI) to support monitoring of cucumber beetles in watermelon fields.
The work focused on assessing whether automated image analysis of yellow
sticky cards can reflect insect activity patterns observed through traditional
manual scouting, and was done in collaboration with David Owens (UD
Extension Specialist) and Nichole Krambeck (Insight Ag Scouting LLC).
Across the 2025 growing season, a comprehensive dataset of sticky card images
was collected from field deployments at UD’s Carvel Research and Education
Center, and at commercial fields scouted by Nichole Krambeck. At the time of
reporting, preliminary quantitative analyses were completed using approximately
15% of the total dataset (≈100 of 625 images). These early analyses indicate that
AI-assisted insect counting is feasible and broadly reflects observed levels of
insect activity.
However, full dataset processing and refinement are ongoing. This
project has established an end-to-end pipeline for AI-assisted pest monitoring,
encompassing field image capture, annotation, cloud-based model deployment,
and quantitative comparison with human counts.