Project Details

[Return to Previous Page]

Automated Smart Trap for detecting and counting pests in agricultural fields

Company: Heather Grab, Assistant Professor, Penn State Entomology

Major(s):
Primary: ME
Secondary: CMPSC
Optional: CMPEN

Non-Disclosure Agreement: NO

Intellectual Property: NO

This project will develop a low cost, field deployable smart sticky card trap that uses computer vision to automatically detect and count thrips on standard yellow sticky cards in near real time. The system will integrate an embedded camera, onboard processing, and a lightweight machine learning model trained to distinguish thrips from non target insects and debris. Images will be collected at regular intervals and processed in situ, with summarized counts transmitted daily via cellular networks to a central server. The trap is designed for use in both research and commercial production settings, including high tunnels and open field systems. Hardware and software choices will prioritize durability, low power consumption, and affordability, with a target cost of less than $100 per unit to support deployment at scale. Automated data collection will reduce labor demands associated with manual card processing while increasing sampling frequency and temporal resolution. By providing timely, spatially explicit information on thrips population dynamics, this system will support more responsive integrated pest management decisions under increasingly variable production conditions driven by changing weather patterns. The project will deliver a validated prototype, annotated image training datasets, and a data pipeline compatible with decision support tools used by growers and extension educators.