# GeoVision Labeler: Zero-Shot Geospatial Classification with Vision and Language Models

arXiv: https://arxiv.org/abs/2505.24340 · HTML: https://mlsanity.com/paper/2505.24340/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 21 of its 32 references that are in the map.
- First posted on arXiv 30 May 2025.

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

- [Learning Transferable Visual Models From Natural Language Supervision](https://mlsanity.com/paper/2103.00020/) (2021)
- [GPT-4 Technical Report](https://mlsanity.com/paper/2303.08774/) (2023)
- [The Llama 3 Herd of Models](https://mlsanity.com/paper/2407.21783/) (2024)
- [Kosmos-2: Grounding Multimodal Large Language Models to the World](https://mlsanity.com/paper/2306.14824/) (2023)
- [Visual Classification via Description from Large Language Models](https://mlsanity.com/paper/2210.07183/) (2022)
- [Why are Visually-Grounded Language Models Bad at Image Classification?](https://mlsanity.com/paper/2405.18415/) (2024)
- [Fine-graind Image Classification via Combining Vision and Language](https://mlsanity.com/paper/1704.02792/) (2017)
- [LLMs for Explainable AI: A Comprehensive Survey](https://mlsanity.com/paper/2504.00125/) (2025)
- [LaFTer: Label-Free Tuning of Zero-shot Classifier using Language and Unlabeled Image Collections](https://mlsanity.com/paper/2305.18287/) (2023)
- [In-domain representation learning for remote sensing](https://mlsanity.com/paper/1911.06721/) (2019)
- [Remote Sensing Image Scene Classification with Self-Supervised Paradigm under Limited Labeled Samples](https://mlsanity.com/paper/2010.00882/) (2020)
- [Revisiting pre-trained remote sensing model benchmarks: resizing and normalization matters](https://mlsanity.com/paper/2305.13456/) (2023)
- [Fine-grained Recognition: Accounting for Subtle Differences between Similar Classes](https://mlsanity.com/paper/1912.06842/) (2019)
- [RS-MetaNet: Deep meta metric learning for few-shot remote sensing scene classification](https://mlsanity.com/paper/2009.13364/) (2020)
- [LangXAI: Integrating Large Vision Models for Generating Textual Explanations to Enhance Explainability in Visual Perception Tasks](https://mlsanity.com/paper/2402.12525/) (2024)
- [Disentangling Semantic-to-visual Confusion for Zero-shot Learning](https://mlsanity.com/paper/2106.08605/) (2021)
- [Taxonomy-Aware Evaluation of Vision-Language Models](https://mlsanity.com/paper/2504.05457/) (2025)
- [The SpaceNet Multi-Temporal Urban Development Challenge](https://mlsanity.com/paper/2102.11958/) (2021)
- [Subset Feature Learning for Fine-Grained Category Classification](https://mlsanity.com/paper/1505.02269/) (2015)
- [Leverage Weakly Annotation to Pixel-wise Annotation via Zero-shot Segment Anything Model for Molecular-empowered Learning](https://mlsanity.com/paper/2308.05785/) (2023)
- [Cost-effective Land Cover Classification for Remote Sensing Images](https://mlsanity.com/paper/2107.12016/) (2021)

## Abstract (arXiv)

Classifying geospatial imagery remains a major bottleneck for applications such as disaster response and land-use monitoring-particularly in regions where annotated data is scarce or unavailable. Existing tools (e.g., RS-CLIP) that claim zero-shot classification capabilities for satellite imagery nonetheless rely on task-specific pretraining and adaptation to reach competitive performance. We introduce GeoVision Labeler (GVL), a strictly zero-shot classification framework: a vision Large Language Model (vLLM) generates rich, human-readable image descriptions, which are then mapped to user-defined classes by a conventional Large Language Model (LLM). This modular, and interpretable pipeline enables flexible image classification for a large range of use cases. We evaluated GVL across three benchmarks-SpaceNet v7, UC Merced, and RESISC45. It achieves up to 93.2% zero-shot accuracy on the binary Buildings vs. No Buildings task on SpaceNet v7. For complex multi-class classification tasks (UC Merced, RESISC45), we implemented a recursive LLM-driven clustering to form meta-classes at successive depths, followed by hierarchical classification-first resolving coarse groups, then finer distinctions-to deliver competitive zero-shot performance. GVL is open-sourced at https://github.com/microsoft/geo-vision-labeler to catalyze adoption in real-world geospatial workflows.