Plug-and-Play Grounding of Reasoning in Multimodal Large Language Models

The surge of Multimodal Large Language Models (MLLMs), given their prominent emergent capabilities in instruction following and reasoning, has greatly advanced the field of visual <PRE_TAG>reasoning</POST_TAG>. However, constrained by their non-lossless image tokenization, most MLLMs fall short of comprehensively capturing details of text and objects, especially in high-resolution images. To address this, we propose P2G, a novel framework for plug-and-play grounding of reasoning in MLLMs. Specifically, P2G exploits the tool-usage potential of MLLMs to employ expert agents to achieve on-the-fly grounding to critical visual and textual objects of image, thus achieving deliberate reasoning via multimodal prompting. We further create <PRE_TAG>P2GB</POST_TAG>, a benchmark aimed at assessing MLLMs' ability to understand inter-object relationships and text in challenging high-resolution images. Comprehensive experiments on <PRE_TAG>visual <PRE_TAG>reasoning</POST_TAG> tasks</POST_TAG> demonstrate the superiority of P2G. Noteworthy, P2G achieved comparable performance with GPT-4V on <PRE_TAG>P2GB</POST_TAG>, with a 7B backbone. Our work highlights the potential of plug-and-play grounding of reasoning and opens up a promising alternative beyond model scaling.