An AI-driven image processing technique to simplify the pollen measurement in common ragweed (Ambrosia artemisiifolia L.)

Ambrosia artemisiifolia (common ragweed) is an invasive weed species that significantly impacts agriculture and public health. This study aimed to develop an automated AI-based object detection model using our annotated image recognition dataset for accurate pollen size measurement, focusing on repeatability and variability in pollen size among individuals with distinct morphological characteristics. The model can effectively streamline the traditionally labour-intensive process, achieving rapid, accurate data collection. Roboflow-based image analysis takes only milliseconds, which is significantly faster than traditional approaches, and a high repeatability index demonstrates a valid methodology for pollen analysis. The study suggests a relationship between pollen size variability and plant morphology, suggesting possible trade-offs between growth and reproduction or showing habitat-specific adaptations. Results may create valuable opportunities for plant biology or ecology, for instance, further investigation of plant-pathogen interactions and public health research. This innovative method represents a step forward in efficient pollen analysis and its integration into multidisciplinary studies.