Abstract: Visual Question Answering (VQA) represents a fundamental challenge in multimodal artificial intelligence, requiring a fine-grained understanding of both visual scenes and natural language ...