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 reasoning. 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 P2GB, a benchmark aimed at assessing MLLMs' ability to understand inter-object relationships and text in challenging high-resolution images. Comprehensive experiments on visual reasoning tasks demonstrate the superiority of P2G. Noteworthy, P2G achieved comparable performance with GPT-4V on P2GB, 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.
We present a neural radiance field method for urban-scale semantic and building-level instance segmentation from aerial images by lifting noisy 2D labels to 3D. This is a challenging problem due to two primary reasons. Firstly, objects in urban aerial images exhibit substantial variations in size, including buildings, cars, and roads, which pose a significant challenge for accurate 2D segmentation. Secondly, the 2D labels generated by existing segmentation methods suffer from the multi-view inconsistency problem, especially in the case of aerial images, where each image captures only a small portion of the entire scene. To overcome these limitations, we first introduce a scale-adaptive semantic label fusion strategy that enhances the segmentation of objects of varying sizes by combining labels predicted from different altitudes, harnessing the novel-view synthesis capabilities of NeRF. We then introduce a novel cross-view instance label grouping strategy based on the 3D scene representation to mitigate the multi-view inconsistency problem in the 2D instance labels. Furthermore, we exploit multi-view reconstructed depth priors to improve the geometric quality of the reconstructed radiance field, resulting in enhanced segmentation results. Experiments on multiple real-world urban-scale datasets demonstrate that our approach outperforms existing methods, highlighting its effectiveness.
Multi-modal large language models (MLLMs) have been shown to efficiently integrate natural language with visual information to handle multi-modal tasks. However, MLLMs still face a fundamental limitation of hallucinations, where they tend to generate erroneous or fabricated information. In this paper, we address hallucinations in MLLMs from a novel perspective of representation learning. We first analyzed the representation distribution of textual and visual tokens in MLLM, revealing two important findings: 1) there is a significant gap between textual and visual representations, indicating unsatisfactory cross-modal representation alignment; 2) representations of texts that contain and do not contain hallucinations are entangled, making it challenging to distinguish them. These two observations inspire us with a simple yet effective method to mitigate hallucinations. Specifically, we introduce contrastive learning into MLLMs and use text with hallucination as hard negative examples, naturally bringing representations of non-hallucinative text and visual samples closer while pushing way representations of non-hallucinating and hallucinative text. We evaluate our method quantitatively and qualitatively, showing its effectiveness in reducing hallucination occurrences and improving performance across multiple benchmarks. On the MMhal-Bench benchmark, our method obtains a 34.66% /29.5% improvement over the baseline MiniGPT-4/LLaVA.