Fine-Grained Preference Optimization Improves Spatial Reasoning in VLMs

TL;DR

This paper introduces SpatialReasoner-R1, a model that significantly improves fine-grained spatial reasoning in vision-language models through novel training methods and preference optimization, achieving state-of-the-art results.

Contribution

The paper presents a new model and training framework that enhances spatial reasoning in VLMs, including a Multi-Model Monte Carlo Tree Search and a fine-grained preference optimization technique.

Findings

fDPO improves spatial reasoning performance by up to 9.0%.

SpatialReasoner-R1 outperforms baselines on SpatialRGPT-Bench.

The methods maintain strong performance on general vision-language tasks.

Abstract

Current Vision-Language Models (VLMs) struggle with fine-grained spatial reasoning, particularly when multi-step logic and precise spatial alignment are required. In this work, we introduce SpatialReasoner-R1, a vision-language reasoning model designed to address these limitations. To construct high-quality supervision for spatial reasoning, we design a Multi-Model Monte Carlo Tree Search (M3CTS) method that generates diverse, logically consistent Long Chain-of-Thought (LongCOT) reasoning trajectories. In addition, we propose a fine-grained Direct Preference Optimization (fDPO) method that introduces segment-specific preference granularity for descriptive grounding and logical reasoning, guided by a spatial reward mechanism that evaluates candidate responses based on visual consistency, spatial grounding, and logical coherence. Experimental results demonstrate that fDPO achieves relative performance gains of 4.1% and 9.0% over standard DPO on spatial qualitative and quantitative tasks, respectively. SpatialReasoner-R1, trained with fDPO, sets a new SoTA on SpatialRGPT-Bench, outperforming the strongest baseline by 9.4% in average accuracy, while maintaining competitive performance on general vision-language tasks.