# Reasoning with Pixel-level Precision: QVLM Architecture and SQuID Dataset for Quantitative Geospatial Analytics

arXiv: https://arxiv.org/abs/2601.13401 · HTML: https://mlsanity.com/paper/2601.13401/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Placed in Spatial and embodied reasoning, one of 17 directions in LLM reasoning.
- Builds on 25 of its 35 references that are in the map.
- First posted on arXiv 19 Jan 2026.

## Placed in

- LLM reasoning / [Spatial and embodied reasoning](https://mlsanity.com/lr/spatial-and-embodied-reasoning/)
- Multimodal LLMs / vision-language models / [Embodied and 3D vision-language models](https://mlsanity.com/mllm/embodied-and-3d-vision-language-models/)

## Builds on (most cited first, 25 of 25)

- [Learning Transferable Visual Models From Natural Language Supervision](https://mlsanity.com/paper/2103.00020/) (2021)
- [Qwen3 Technical Report](https://mlsanity.com/paper/2505.09388/) (2025)
- [Visual Instruction Tuning](https://mlsanity.com/paper/2304.08485/) (2023)
- [DINOv3](https://mlsanity.com/paper/2508.10104/) (2025)
- [gpt-oss-120b & gpt-oss-20b Model Card](https://mlsanity.com/paper/2508.10925/) (2025)
- [LISA: Reasoning Segmentation via Large Language Model](https://mlsanity.com/paper/2308.00692/) (2023)
- [SpatialVLM: Endowing Vision-Language Models with Spatial Reasoning Capabilities](https://mlsanity.com/paper/2401.12168/) (2024)
- [When and why vision-language models behave like bags-of-words, and what to do about it?](https://mlsanity.com/paper/2210.01936/) (2022)
- [ViperGPT: Visual Inference via Python Execution for Reasoning](https://mlsanity.com/paper/2303.08128/) (2023)
- [Vision Transformer Adapter for Dense Predictions](https://mlsanity.com/paper/2205.08534/) (2022)
- [Visual Programming: Compositional visual reasoning without training](https://mlsanity.com/paper/2211.11559/) (2022)
- [RemoteCLIP: A Vision Language Foundation Model for Remote Sensing](https://mlsanity.com/paper/2306.11029/) (2023)
- [GeoChat: Grounded Large Vision-Language Model for Remote Sensing](https://mlsanity.com/paper/2311.15826/) (2023)
- [VHM: Versatile and Honest Vision Language Model for Remote Sensing Image Analysis](https://mlsanity.com/paper/2403.20213/) (2024)
- [EarthVQA: Towards Queryable Earth via Relational Reasoning-Based Remote Sensing Visual Question Answering](https://mlsanity.com/paper/2312.12222/) (2023)
- [GEOBench-VLM: Benchmarking Vision-Language Models for Geospatial Tasks](https://mlsanity.com/paper/2411.19325/) (2024)
- [Modular Visual Question Answering via Code Generation](https://mlsanity.com/paper/2306.05392/) (2023)
- [Text2Seg: Remote Sensing Image Semantic Segmentation via Text-Guided Visual Foundation Models](https://mlsanity.com/paper/2304.10597/) (2023)
- [Good at captioning, bad at counting: Benchmarking GPT-4V on Earth observation data](https://mlsanity.com/paper/2401.17600/) (2024)
- [SegEarth-OV: Towards Training-Free Open-Vocabulary Segmentation for Remote Sensing Images](https://mlsanity.com/paper/2410.01768/) (2024)
- [HRVQA: A Visual Question Answering Benchmark for High-Resolution Aerial Images](https://mlsanity.com/paper/2301.09460/) (2023)
- [ConMe: Rethinking Evaluation of Compositional Reasoning for Modern VLMs](https://mlsanity.com/paper/2406.08164/) (2024)
- [Lost in Embeddings: Information Loss in Vision-Language Models](https://mlsanity.com/paper/2509.11986/) (2025)
- [RSVLM-QA: A Benchmark Dataset for Remote Sensing Vision Language Model-based Question Answering](https://mlsanity.com/paper/2508.07918/) (2025)
- [Aerial Imagery for Roof Segmentation: A Large-Scale Dataset towards Automatic Mapping of Buildings](https://mlsanity.com/paper/1807.09532/) (2018)

## Built on by (most cited first, 1 of 1)

- [ReVA: A Scene-Centric Dataset Beyond Repetition for Remote Sensing Video Question Answering](https://mlsanity.com/paper/2609.35507/) (2026)

## Abstract (arXiv)

Current Vision-Language Models (VLMs) fail at quantitative spatial reasoning because their architectures destroy pixel-level information required for counting and measurements. Vision encoders compress images through patch embeddings, reducing spatial indexing and losing the precise pixel-level tracking required for accurate counting. We present two contributions to address this fundamental limitation. First, we introduce SQuID (Satellite Quantitative Intelligence Dataset), a benchmark of 2,000 satellite image Question-Answer pairs with both numerical range and categorical answers, designed to evaluate quantitative spatial reasoning. The dataset spans three difficulty tiers with annotations automatically generated from human labels and their learned variability. Second, we propose QVLM (Quantitative Vision-Language Model), a code-generation architecture that maintains pixel precision by decoupling language understanding from visual analysis. Instead of encoding images into embeddings, QVLM generates executable code that first calls a segmentation model to obtain pixel-level masks, then operates directly on these masks, preserving spatial indexing throughout the reasoning process. Our experiments show that QVLM using GPT-5 as coder achieves 42.0% accuracy on SQuID compared to 28.1% for a VLM prompted with image-question pairs. Our work reveals that, for quantitative spatial reasoning, architectural decoupling enables better accuracy on quantitative tasks.