Fine-Grained Retrieval-Augmented Generation for Visual Question Answering (2502.20964v2)

Abstract: Visual Question Answering (VQA) focuses on providing answers to natural language questions by utilizing information from images. Although cutting-edge multimodal LLMs (MLLMs) such as GPT-4o achieve strong performance on VQA tasks, they frequently fall short in accessing domain-specific or the latest knowledge. To mitigate this issue, retrieval-augmented generation (RAG) leveraging external knowledge bases (KBs), referred to as KB-VQA, emerges as a promising approach. Nevertheless, conventional unimodal retrieval techniques, which translate images into textual descriptions, often result in the loss of critical visual details. This study presents fine-grained knowledge units, which merge textual snippets with entity images stored in vector databases. Furthermore, we introduce a knowledge unit retrieval-augmented generation framework (KU-RAG) that integrates fine-grained retrieval with MLLMs. The proposed KU-RAG framework ensures precise retrieval of relevant knowledge and enhances reasoning capabilities through a knowledge correction chain. Experimental findings demonstrate that our approach significantly boosts the performance of leading KB-VQA methods, achieving an average improvement of approximately 3% and up to 11% in the best case.