Unleashing Spatial Reasoning in Multimodal Large Language Models via Textual Representation Guided Reasoning
Fuente:
arXiv
Guardado en:
| Autores principales: | , , , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2026
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866913046931177472 |
|---|---|
| author | Hua, Jiacheng Yin, Yishu Wu, Yuhang Wang, Tai Huang, Yifei Liu, Miao |
| author_facet | Hua, Jiacheng Yin, Yishu Wu, Yuhang Wang, Tai Huang, Yifei Liu, Miao |
| contents | Existing Multimodal Large Language Models (MLLMs) struggle with 3D spatial reasoning, as they fail to construct structured abstractions of the 3D environment depicted in video inputs. To bridge this gap, drawing inspiration from cognitive theories of allocentric spatial reasoning, we investigate how to enable MLLMs to model and reason over text-based spatial representations of video. Specifically, we introduce Textual Representation of Allocentric Context from Egocentric Video (TRACE), a prompting method that induces MLLMs to generate text-based representations of 3D environments as intermediate reasoning traces for more accurate spatial question answering. TRACE encodes meta-context, camera trajectories, and detailed object entities to support structured spatial reasoning over egocentric videos. Extensive experiments on VSI-Bench and OST-Bench demonstrate that TRACE yields notable and consistent improvements over prior prompting strategies across a diverse range of MLLM backbones, spanning different parameter scales and training schemas. We further present ablation studies to validate our design choices, along with detailed analyses that probe the bottlenecks of 3D spatial reasoning in MLLMs. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2603_23404 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Unleashing Spatial Reasoning in Multimodal Large Language Models via Textual Representation Guided Reasoning Hua, Jiacheng Yin, Yishu Wu, Yuhang Wang, Tai Huang, Yifei Liu, Miao Computer Vision and Pattern Recognition Computation and Language Existing Multimodal Large Language Models (MLLMs) struggle with 3D spatial reasoning, as they fail to construct structured abstractions of the 3D environment depicted in video inputs. To bridge this gap, drawing inspiration from cognitive theories of allocentric spatial reasoning, we investigate how to enable MLLMs to model and reason over text-based spatial representations of video. Specifically, we introduce Textual Representation of Allocentric Context from Egocentric Video (TRACE), a prompting method that induces MLLMs to generate text-based representations of 3D environments as intermediate reasoning traces for more accurate spatial question answering. TRACE encodes meta-context, camera trajectories, and detailed object entities to support structured spatial reasoning over egocentric videos. Extensive experiments on VSI-Bench and OST-Bench demonstrate that TRACE yields notable and consistent improvements over prior prompting strategies across a diverse range of MLLM backbones, spanning different parameter scales and training schemas. We further present ablation studies to validate our design choices, along with detailed analyses that probe the bottlenecks of 3D spatial reasoning in MLLMs. |
| title | Unleashing Spatial Reasoning in Multimodal Large Language Models via Textual Representation Guided Reasoning |
| topic | Computer Vision and Pattern Recognition Computation and Language |
| url | https://arxiv.org/abs/2603.23404 |