Honey authentication using AI-based pollen analysis: a UK review

Purpose
Honey authentication is essential for ensuring food quality and detecting adulteration, especially in the UK, where large amounts of honey are imported and often lacks consistent production and quality standards. The UK honey market faces a significant challenge in verifying its botanical origins, which is crucial for detecting adulteration. This study aims to summarise the impact of recent advancements in honey pollen analysis that use AI while addressing a critical gap in the existing literature within the UK context.

Design/methodology/approach
Drawing upon a systematic review, the paper incorporates a thematic analysis of UK regulations, pollen dataset collection, dataset preparation techniques, different detection and classification models, technical challenges within honey authentication and its importance for the UK.

Findings
This study examines research worldwide, including publicly available pollen datasets, robust pollen detection systems, image processing techniques, deep learning architectures and traditional machine learning algorithms, showing promising advancements in the field, with some accuracies exceeding 95%. The UK lacks research on honey authentication, showcasing a gap between the honey market challenges and technological advancements.

Originality/value
Unlike existing reviews that focus primarily on technical aspects, this study integrates current research in AI advancements within the context of UK honey regulations and market challenges. It highlights technological limitations and regulatory barriers while bridging the gap between innovation and policy. The paper offers recommendations for future research to enhance honey authentication in the UK.