Enhancing MLLM Spatial Understanding via Active 3D Scene Exploration for Multi-Perspective Reasoning

TL;DR

This paper introduces a training-free framework that improves 3D spatial reasoning in Multimodal Large Language Models by reconstructing 3D scenes and synthesizing novel views for better perspective understanding.

Contribution

It presents a novel, training-free approach that combines 3D reconstruction and view synthesis to enhance spatial reasoning in MLLMs, outperforming existing models.

Findings

Outperforms GPT-5.2 and Gemini-2.5-Flash on 3DSRBench and Rel3D benchmarks.

Significantly improves spatial comprehension in MLLMs.

Utilizes explicit 3D reconstruction and external knowledge for viewpoint synthesis.

Abstract

Although Multimodal Large Language Models have achieved remarkable progress, they still struggle with complex 3D spatial reasoning due to the reliance on 2D visual priors. Existing approaches typically mitigate this limitation either through computationally expensive post-training procedures on limited 3D datasets or through rigid tool-calling mechanisms that lack explicit geometric understanding and viewpoint flexibility. To address these challenges, we propose a \textit{training-free} framework that introduces a Visual Chain-of-Thought mechanism grounded in explicit 3D reconstruction. The proposed pipeline first reconstructs a high-fidelity 3D mesh from a single image using MLLM-guided keyword extraction and mask generation at multiple granularities. Subsequently, the framework leverages an external knowledge base to iteratively compute optimal camera extrinsic parameters and synthesize novel views, thereby emulating human perspective-taking. Extensive experiments demonstrate that the proposed approach significantly enhances spatial comprehension. Specifically, the framework outperforms specialized spatial models and general-purpose MLLMs, including \textit{GPT-5.2} and \textit{Gemini-2.5-Flash}, on major benchmarks such as 3DSRBench and Rel3D.