CityCube: Benchmarking Cross-view Spatial Reasoning on Vision-Language Models in Urban Environments

TL;DR

CityCube is a new benchmark designed to evaluate vision-language models' ability to perform cross-view spatial reasoning in complex urban environments, revealing significant gaps between current models and human performance.

Contribution

It introduces a comprehensive urban-focused benchmark with diverse viewpoints and annotated QA pairs, addressing a gap in existing spatial reasoning benchmarks.

Findings

Current VLMs perform significantly worse than humans in urban spatial reasoning.

Small-scale fine-tuned VLMs outperform large-scale models on this benchmark.

There is a fundamental cognitive gap between VLMs and human spatial reasoning.

Abstract

Cross-view spatial reasoning is essential for embodied AI, underpinning spatial understanding, mental simulation and planning in complex environments. Existing benchmarks primarily emphasize indoor or street settings, overlooking the unique challenges of open-ended urban spaces characterized by rich semantics, complex geometries, and view variations. To address this, we introduce CityCube, a systematic benchmark designed to probe cross-view reasoning capabilities of current VLMs in urban settings. CityCube integrates four viewpoint dynamics to mimic camera movements and spans a wide spectrum of perspectives from multiple platforms, e.g., vehicles, drones and satellites. For a comprehensive assessment, it features 5,022 meticulously annotated multi-view QA pairs categorized into five cognitive dimensions and three spatial relation expressions. A comprehensive evaluation of 33 VLMs reveals a significant performance disparity with humans: even large-scale models struggle to exceed 54.1% accuracy, remaining 34.2% below human performance. By contrast, small-scale fine-tuned VLMs achieve over 60.0% accuracy, highlighting the necessity of our benchmark. Further analyses indicate the task correlations and fundamental cognitive disparity between VLMs and human-like reasoning.