CrossView Suite: Harnessing Cross-view Spatial Intelligence of MLLMs with Dataset, Model and Benchmark

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Wang, Wei, Yuan, Yuqian, Lin, Tianwei, Zhang, Wenqiao, Tang, Siliang, Xiao, Jun, Zhuang, Yueting
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866910232789123072
author Wang, Wei
Yuan, Yuqian
Lin, Tianwei
Zhang, Wenqiao
Tang, Siliang
Xiao, Jun
Zhuang, Yueting
author_facet Wang, Wei
Yuan, Yuqian
Lin, Tianwei
Zhang, Wenqiao
Tang, Siliang
Xiao, Jun
Zhuang, Yueting
contents Spatial intelligence requires multimodal large language models (MLLMs) to move beyond single-view perception and reason consistently about objects, visibility, geometry, and interactions across multiple viewpoints. However, progress in cross-view reasoning remains limited by three major gaps: the scarcity of large-scale well-annotated training data, the lack of comprehensive benchmarks for systematic evaluation, and the absence of explicit alignment mechanisms that establish object-level consistency across views. To address these gaps, we thoroughly develop CrossView Suite across three coordinated components: CrossViewSet, CrossViewBench, and CrossViewer. Firstly, we introduce a multi-agent data engine to meticulously curate a large-scale, high-quality cross-view instruction dataset, termed CrossViewSet, covering 17 fine-grained task types with 1.6M samples. Second, we meticulously create a scene-disjoint CrossViewBench to comprehensively assess the cross-view spatial understanding capability of an MLLM, evaluating it across various aspects. Finally, we propose CrossViewer, a progressive three-stage framework for cross-view spatial reasoning in MLLMs, following a Perception -> Alignment -> Reasoning paradigm. Our method equips an adaptive spatial region tokenizer to capture fine-grained object representations, and then aligns the multi-view objects explicitly, and thus fuses aligned features for boosting the cross-view inference capacity for MLLMs. Extensive experiments and analyses show that large-scale training data, systematic evaluation, and explicit cross-view alignment are all critical for advancing MLLMs from single-view perception toward real-world spatial intelligence. The project page is available at https://github.com/Thinkirin/Crossview-Suite.
format Preprint
id arxiv_https___arxiv_org_abs_2605_18621
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle CrossView Suite: Harnessing Cross-view Spatial Intelligence of MLLMs with Dataset, Model and Benchmark
Wang, Wei
Yuan, Yuqian
Lin, Tianwei
Zhang, Wenqiao
Tang, Siliang
Xiao, Jun
Zhuang, Yueting
Computer Vision and Pattern Recognition
Artificial Intelligence
Spatial intelligence requires multimodal large language models (MLLMs) to move beyond single-view perception and reason consistently about objects, visibility, geometry, and interactions across multiple viewpoints. However, progress in cross-view reasoning remains limited by three major gaps: the scarcity of large-scale well-annotated training data, the lack of comprehensive benchmarks for systematic evaluation, and the absence of explicit alignment mechanisms that establish object-level consistency across views. To address these gaps, we thoroughly develop CrossView Suite across three coordinated components: CrossViewSet, CrossViewBench, and CrossViewer. Firstly, we introduce a multi-agent data engine to meticulously curate a large-scale, high-quality cross-view instruction dataset, termed CrossViewSet, covering 17 fine-grained task types with 1.6M samples. Second, we meticulously create a scene-disjoint CrossViewBench to comprehensively assess the cross-view spatial understanding capability of an MLLM, evaluating it across various aspects. Finally, we propose CrossViewer, a progressive three-stage framework for cross-view spatial reasoning in MLLMs, following a Perception -> Alignment -> Reasoning paradigm. Our method equips an adaptive spatial region tokenizer to capture fine-grained object representations, and then aligns the multi-view objects explicitly, and thus fuses aligned features for boosting the cross-view inference capacity for MLLMs. Extensive experiments and analyses show that large-scale training data, systematic evaluation, and explicit cross-view alignment are all critical for advancing MLLMs from single-view perception toward real-world spatial intelligence. The project page is available at https://github.com/Thinkirin/Crossview-Suite.
title CrossView Suite: Harnessing Cross-view Spatial Intelligence of MLLMs with Dataset, Model and Benchmark
topic Computer Vision and Pattern Recognition
Artificial Intelligence
url https://arxiv.org/abs/2605.18621