Guardado en:
| Autores principales: | , , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2026
|
| Materias: | |
| Acceso en línea: | https://arxiv.org/abs/2604.21190 |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866914512489152512 |
|---|---|
| author | Hwang, Chan Yeong Choi, Miso On, Sunghyun Kim, Jinkyu Lee, Jungbeom |
| author_facet | Hwang, Chan Yeong Choi, Miso On, Sunghyun Kim, Jinkyu Lee, Jungbeom |
| contents | Understanding visual scenes requires not only recognizing objects but also reasoning about their spatial relationships. Unlike general vision-language tasks, spatial reasoning requires integrating multiple inductive biases, such as 2D appearance cues, depth signals, and geometric constraints, whose reliability varies across contexts. This suggests that effective spatial reasoning requires \emph{spatial adaptability}: the ability to flexibly coordinate different reasoning strategies depending on the input. However, most existing approaches rely on a single reasoning pipeline that implicitly learns a fixed spatial prior, limiting their ability to adapt under distribution changes. Multi-agent systems offer a promising alternative by aggregating diverse reasoning trajectories, but prior attempts in spatial reasoning primarily employ homogeneous agents, restricting the diversity of inductive biases they can leverage. In this work, we introduce SpatiO, a heterogeneous multi-agent framework for spatial reasoning that coordinates multiple vision-language specialists with complementary inductive biases. To enable effective collaboration, we propose Test-Time Orchestration (TTO), an optimization mechanism that dynamically evaluates and reweights agents based on their observed reliability during inference, without modifying model parameters. Extensive experiments on diverse spatial reasoning benchmarks, including 3DSRBench, STVQA-7k, CV-Bench, and Omni3D-Bench, demonstrate that SpatiO consistently improves spatial reasoning performance over both closed-source and open-source baselines. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2604_21190 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | SpatiO: Adaptive Test-Time Orchestration of Vision-Language Agents for Spatial Reasoning Hwang, Chan Yeong Choi, Miso On, Sunghyun Kim, Jinkyu Lee, Jungbeom Computer Vision and Pattern Recognition Understanding visual scenes requires not only recognizing objects but also reasoning about their spatial relationships. Unlike general vision-language tasks, spatial reasoning requires integrating multiple inductive biases, such as 2D appearance cues, depth signals, and geometric constraints, whose reliability varies across contexts. This suggests that effective spatial reasoning requires \emph{spatial adaptability}: the ability to flexibly coordinate different reasoning strategies depending on the input. However, most existing approaches rely on a single reasoning pipeline that implicitly learns a fixed spatial prior, limiting their ability to adapt under distribution changes. Multi-agent systems offer a promising alternative by aggregating diverse reasoning trajectories, but prior attempts in spatial reasoning primarily employ homogeneous agents, restricting the diversity of inductive biases they can leverage. In this work, we introduce SpatiO, a heterogeneous multi-agent framework for spatial reasoning that coordinates multiple vision-language specialists with complementary inductive biases. To enable effective collaboration, we propose Test-Time Orchestration (TTO), an optimization mechanism that dynamically evaluates and reweights agents based on their observed reliability during inference, without modifying model parameters. Extensive experiments on diverse spatial reasoning benchmarks, including 3DSRBench, STVQA-7k, CV-Bench, and Omni3D-Bench, demonstrate that SpatiO consistently improves spatial reasoning performance over both closed-source and open-source baselines. |
| title | SpatiO: Adaptive Test-Time Orchestration of Vision-Language Agents for Spatial Reasoning |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2604.21190 |