# m2sv: A Scalable Benchmark for Map-to-Street-View Spatial Reasoning

arXiv: https://arxiv.org/abs/2601.19099 · HTML: https://mlsanity.com/paper/2601.19099/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in Spatial and embodied reasoning, one of 17 directions in LLM reasoning.
- Builds on 21 of its 30 references that are in the map.
- First posted on arXiv 27 Jan 2026.

## Placed in

- LLM reasoning / [Spatial and embodied reasoning](https://mlsanity.com/lr/spatial-and-embodied-reasoning/)

## Builds on (most cited first, 21 of 21)

- [MathVista: Evaluating Mathematical Reasoning of Foundation Models in Visual Contexts](https://mlsanity.com/paper/2310.02255/) (2023)
- [FrontierMath: A Benchmark for Evaluating Advanced Mathematical Reasoning in AI](https://mlsanity.com/paper/2411.04872/) (2024)
- [VIGOR: Cross-View Image Geo-localization beyond One-to-one Retrieval](https://mlsanity.com/paper/2011.12172/) (2020)
- [Wide-Area Image Geolocalization with Aerial Reference Imagery](https://mlsanity.com/paper/1510.03743/) (2015)
- [Lending Orientation to Neural Networks for Cross-view Geo-localization](https://mlsanity.com/paper/1903.12351/) (2019)
- [University-1652: A Multi-view Multi-source Benchmark for Drone-based Geo-localization](https://mlsanity.com/paper/2002.12186/) (2020)
- [DynaMath: A Dynamic Visual Benchmark for Evaluating Mathematical Reasoning Robustness of Vision Language Models](https://mlsanity.com/paper/2411.00836/) (2024)
- [Where am I looking at? Joint Location and Orientation Estimation by Cross-View Matching](https://mlsanity.com/paper/2005.03860/) (2020)
- [PlaNet - Photo Geolocation with Convolutional Neural Networks](https://mlsanity.com/paper/1602.05314/) (2016)
- [Mapping global dynamics of benchmark creation and saturation in artificial intelligence](https://mlsanity.com/paper/2203.04592/) (2022)
- [Spatial Reasoning with Vision-Language Models in Ego-Centric Multi-View Scenes](https://mlsanity.com/paper/2509.06266/) (2025)
- [PIGEON: Predicting Image Geolocations](https://mlsanity.com/paper/2307.05845/) (2023)
- [Revisiting Street-to-Aerial View Image Geo-localization and Orientation Estimation](https://mlsanity.com/paper/2005.11592/) (2020)
- [UrBench: A Comprehensive Benchmark for Evaluating Large Multimodal Models in Multi-View Urban Scenarios](https://mlsanity.com/paper/2408.17267/) (2024)
- [CV-Cities: Advancing Cross-View Geo-Localization in Global Cities](https://mlsanity.com/paper/2411.12431/) (2024)
- [MapQA: Open-domain Geospatial Question Answering on Map Data](https://mlsanity.com/paper/2503.07871/) (2025)
- [Self-Contradictory Reasoning Evaluation and Detection](https://mlsanity.com/paper/2311.09603/) (2023)
- [CityCube: Benchmarking Cross-view Spatial Reasoning on Vision-Language Models in Urban Environments](https://mlsanity.com/paper/2601.14339/) (2026)
- [VLMs have Tunnel Vision: Evaluating Nonlocal Visual Reasoning in Leading VLMs](https://mlsanity.com/paper/2507.13361/) (2025)
- [SpaGBOL: Spatial-Graph-Based Orientated Localisation](https://mlsanity.com/paper/2409.15514/) (2024)
- [CoNav: Collaborative Cross-Modal Reasoning for Embodied Navigation](https://mlsanity.com/paper/2505.16663/) (2025)

## Abstract (arXiv)

Vision--language models (VLMs) achieve strong performance on many multimodal benchmarks but remain brittle on spatial reasoning tasks that require aligning abstract overhead representations with egocentric views. We introduce m2sv, a scalable benchmark for map-to-street-view spatial reasoning that asks models to infer camera viewing direction by aligning a north-up overhead map with a Street View image captured at the same real-world intersection. We release m2sv-20k, a geographically diverse benchmark with controlled ambiguity, along with m2sv-sft-11k, a curated set of structured reasoning traces for supervised fine-tuning. Despite strong performance on existing multimodal benchmarks, the best evaluated VLM achieves only 65.2% accuracy on m2sv, below human annotators who reach 72.0% on average (and 95% for an expert) with strong inter-annotator agreement (kappa up to 0.76). While supervised fine-tuning and reinforcement learning yield consistent gains, cross-benchmark evaluations reveal limited transfer. Beyond aggregate accuracy, we systematically analyze difficulty in map-to-street-view reasoning using both structural signals and human effort, and conduct an extensive failure analysis of adapted open models. Our findings highlight persistent gaps in geometric alignment, evidence aggregation, and reasoning consistency, motivating future work on grounded spatial reasoning across viewpoints.