3D-Aware Vision-Language Models Fine-Tuning with Geometric Distillation

TL;DR

This paper introduces Geometric Distillation, a lightweight fine-tuning method that enhances pretrained vision-language models with 3D spatial understanding by distilling geometric cues from existing 3D models, without altering the original architecture.

Contribution

It presents a novel, annotation-free framework for injecting 3D geometric knowledge into VLMs, improving 3D reasoning capabilities efficiently and compatibly with existing models.

Findings

Outperforms prior methods on 3D vision-language benchmarks.

Achieves better 3D spatial reasoning with lower computational cost.

Effectively integrates geometric cues without architecture modifications.

Abstract

Vision-Language Models (VLMs) have shown remarkable performance on diverse visual and linguistic tasks, yet they remain fundamentally limited in their understanding of 3D spatial structures. We propose Geometric Distillation, a lightweight, annotation-free fine-tuning framework that injects human-inspired geometric cues into pretrained VLMs without modifying their architecture. By distilling (1) sparse correspondences, (2) relative depth relations, and (3) dense cost volumes from off-the-shelf 3D foundation models (e.g., MASt3R, VGGT), our method shapes representations to be geometry-aware while remaining compatible with natural image-text inputs. Through extensive evaluations on 3D vision-language reasoning and 3D perception benchmarks, our method consistently outperforms prior approaches, achieving improved 3D spatial reasoning with significantly lower computational cost. Our work demonstrates a scalable and efficient path to bridge 2D-trained VLMs with 3D understanding, opening up wider use in spatially grounded multimodal tasks.