The development of the PSYLLid platform was driven by the urgent need for more efficient monitoring of Asian citrus psyllids in citrus groves. Current surveillance methods require deploying hundreds of yellow sticky cards across trees, retrieving them at regular intervals, and examining each card under magnification to identify any trapped psyllids. This process is highly labor-intensive and demands careful inspection to distinguish one or a few adult psyllids from thousands of other insects captured on each trap. Moreover, the time lag between field collection and laboratory analysis can extend from weeks to months. During this delay, psyllid populations may become established and spread within the grove or to surrounding areas, increasing the risk of infestation and disease transmission.
Ideally, sticky cards should be examined at 1–2-week intervals to enable rapid detection of psyllid presence. To address the limitations of manual inspection and advance monitoring efforts, the Tree Nut and Fruit Entomology Laboratory, in collaboration with the Bugwood Center for Invasive Species and Ecosystem Health at the University of Georgia, developed a computer vision platform powered by a deep learning model. The system is designed to automatically detect and count adult Asian citrus psyllids on yellow sticky cards.
The platform utilizes a Ultralytics YOLO 11 architecture, achieving a precision of 96.9% (mAP@50) and a recall of 92.4%. The model processes each image in approximately 72 milliseconds, enabling rapid analysis. This tool significantly improves the speed and accuracy of Asian citrus psyllid monitoring, providing growers, county agents, and other stakeholders with a more efficient method for evaluating sticky card traps.