Agri-VLG: Robust Identification of Counterfeit Agricultural Products Using Few-Shot Annotated Data and Open-Source Knowledge
Agri-VLG is proposed, a robust framework that synergizes few-shot annotated data with open-source knowledge for precise identification and significantly outperforms state-of-the-art methods, proving that graph-based refinement and external knowledge-guided retrieval are highly effective for detecting counterfeit produc...