VietMEAgent: Culturally-Aware Few-Shot Multimodal Explanation for Vietnamese Visual Question Answering

VietMEAgent represents a significant advancement in Visual Question Answering (VQA) systems, particularly in addressing the cultural knowledge gap that has historically limited AI's effectiveness in understanding culturally specific content. By integrating a cultural object detection backbone with a structured program generation layer, VietMEAgent not only improves answer prediction but also enhances the interpretability of AI responses. The development of a Vietnamese Cultural VQA dataset further supports this initiative, providing a rich source of culturally relevant information. This dataset, along with the dual-modality explanation module, allows the system to deliver transparent explanations that combine visual evidence with human-readable textual rationales. As AI continues to evolve, the emphasis on culturally aware systems like VietMEAgent is crucial for fostering trust and understanding among users, ensuring that AI technologies are inclusive and representative of diverse cult…