Saved in:
Bibliographic Details
Main Authors: Huang, Xiaohu, Wu, Jingjing, Xie, Qunyi, Han, Kai
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2506.01946
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866912752752132096
author Huang, Xiaohu
Wu, Jingjing
Xie, Qunyi
Han, Kai
author_facet Huang, Xiaohu
Wu, Jingjing
Xie, Qunyi
Han, Kai
contents Recent advances in scene understanding have leveraged multimodal large language models (MLLMs) for 3D reasoning by capitalizing on their strong 2D pretraining. However, the lack of explicit 3D data during MLLM pretraining limits 3D representation capability. In this paper, we investigate the 3D-awareness of MLLMs by evaluating multi-view correspondence and reveal a strong positive correlation between the quality of 3D-aware representation and downstream task performance. Motivated by this, we propose 3DRS, a framework that enhances MLLM 3D representation learning by introducing supervision from pretrained 3D foundation models. Our approach aligns MLLM visual features with rich 3D knowledge distilled from 3D models, effectively improving scene understanding. Extensive experiments across multiple benchmarks and MLLMs -- including visual grounding, captioning, and question answering -- demonstrate consistent performance gains. Project page: https://visual-ai.github.io/3drs
format Preprint
id arxiv_https___arxiv_org_abs_2506_01946
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle 3DRS: MLLMs Need 3D-Aware Representation Supervision for Scene Understanding
Huang, Xiaohu
Wu, Jingjing
Xie, Qunyi
Han, Kai
Computer Vision and Pattern Recognition
Recent advances in scene understanding have leveraged multimodal large language models (MLLMs) for 3D reasoning by capitalizing on their strong 2D pretraining. However, the lack of explicit 3D data during MLLM pretraining limits 3D representation capability. In this paper, we investigate the 3D-awareness of MLLMs by evaluating multi-view correspondence and reveal a strong positive correlation between the quality of 3D-aware representation and downstream task performance. Motivated by this, we propose 3DRS, a framework that enhances MLLM 3D representation learning by introducing supervision from pretrained 3D foundation models. Our approach aligns MLLM visual features with rich 3D knowledge distilled from 3D models, effectively improving scene understanding. Extensive experiments across multiple benchmarks and MLLMs -- including visual grounding, captioning, and question answering -- demonstrate consistent performance gains. Project page: https://visual-ai.github.io/3drs
title 3DRS: MLLMs Need 3D-Aware Representation Supervision for Scene Understanding
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2506.01946