Classification of images of bee pollen according to their producers

Abstract

The food industry is witnessing a growing interest in pollen due to its nutritional and energy composition. Consumers of bee pollen are increasingly eager to learn about the origins of the products they purchase. Establishing the geographical origin and the producer of pollen can enhance the product’s value and meet consumer demands for transparency in the supply chain. This article presents a novel approach for the classification of images of bee pollen according to their producers using digital images and machine learning. The study focuses on pollen collected from various beekeepers in the Boyacá region of Colombia. A standardized image acquisition process was employed to capture macroscopic images of the pollen samples. These images were then analyzed to extract color information, and machine learning models were trained to predict the producer of the pollen based on its color characteristics. The results demonstrate that the proposed approach can effectively determine the producer of pollen samples based on their color information. The model achieved an accuracy rate of 85% in associating pollen samples with their respective beekeepers. This outcome has significant implications for traceability and transparency in the bee pollen industry, offering a cost-effective and accessible method to verify the product’s origin.