SQuID: Satellite Quantitative Intelligence Dataset

A comprehensive benchmark for evaluating quantitative spatial reasoning in Vision-Language Models using satellite imagery.

Related Resources

• Code repository: https://github.com/PeterAMassih/qvlm-squid
• Paper (arXiv): https://arxiv.org/abs/2601.13401

Dataset Overview

• 2000 questions testing spatial reasoning on satellite imagery
• 587 unique images across four datasets
• 1950 auto-labeled questions from segmentation masks (DeepGlobe, EarthVQA, Solar Panels)
• 50 human-annotated questions from NAIP imagery with consensus answers
• 1577 questions include human-agreement ranges for numeric answers
• 3 difficulty tiers: Basic (710), Spatial (616), Complex (674)
• 3 resolution levels: 0.3m, 0.5m, 1.0m GSD

Human Annotation & Agreement Methodology

Human Annotation Process

• 50 questions on NAIP 1.0m GSD imagery were annotated by humans
• 10 annotators per question resulting in 500 total annotations
• Answer aggregation:
  • Numeric questions: Used MEDIAN of all responses for robustness
  • Categorical questions (connected/fragmented): Used MAJORITY voting
  • Binary questions: Converted yes/no to 1/0 and used majority

Human Agreement Quantification

From the 500 human annotations, we computed the Mean Median Absolute Deviation (MAD) for each question type:

• Percentage questions: MAD = ±1.74 percentage points
• Proximity questions: MAD = ±2.25 percentage points
• Count questions: Normalized MADc = 0.19 (proportional to count magnitude)

For count questions, we use a normalized MAD (MADc) that makes the acceptable range proportional to the count value:

MADc = median(|Xi - median(X)|) / median(X) = 0.19

Acceptable Range Calculation

These MAD values were applied to ALL numeric questions in the benchmark to define acceptable ranges:

import math

# For percentage questions (absolute deviation)
if question_type == 'percentage':
    lower = max(0.0, answer - 1.74)
    upper = min(100.0, answer + 1.74)
    
# For count questions (proportional deviation)
# range(C) = [C - max(1, C × MADc), C + max(1, C × MADc)]
elif question_type in ['count', 'building_proximity', 'building_flood_risk', 
                        'building_fire_risk', 'connectivity']:
    MADc = 0.19
    dr = max(1, answer * MADc)  # At least ±1 deviation
    lower = max(0, math.floor(answer - dr))
    upper = math.ceil(answer + dr)
    
# For proximity percentage questions (absolute deviation)
elif 'within' in question and 'm of' in question:
    lower = max(0.0, answer - 2.25)
    upper = min(100.0, answer + 2.25)

Example count ranges with MADc = 0.19:

• C=5 → range [4, 7]
• C=10 → range [8, 12]
• C=50 → range [40, 60]
• C=100 → range [81, 120]

Special cases:

• Zero values have no range (exact match required)
• Binary/fragmentation questions have no range (exact match)
• Ranges are capped at valid bounds (0-100 for percentages, ≥0 for counts)

Question Types

The benchmark includes 24 distinct question types organized into three tiers:

Tier 1: Basic Questions (710 questions)

• percentage: Coverage percentage of a land use class
• count: Number of separate regions or objects
• size: Area measurements of regions
• total_area: Total area covered by a class
• binary_comparison: Comparing quantities between two classes
• binary_presence: Checking if a class exists
• binary_threshold: Testing if values exceed thresholds
• binary_multiple: Checking for multiple instances

Tier 2: Spatial Questions (616 questions)

• proximity_percentage: Percentage of one class near another
• proximity_area: Area of one class near another
• binary_proximity: Presence of one class near another
• building_proximity: Number of buildings near other features
• building_flood_risk: Buildings at flood risk (near water)
• building_fire_risk: Buildings at fire risk (near forest)
• connectivity: Counting isolated patches by size
• fragmentation: Assessing if regions are connected or fragmented
• power_calculation: Calculating solar panel power output

Tier 3: Complex Questions (674 questions)

• complex_multi_condition: Areas meeting multiple spatial criteria
• complex_urban_flood_risk: Urban areas at flood risk (near water)
• complex_urban_fire_risk: Urban areas at fire risk (near forest)
• complex_agriculture_water_access: Agricultural land with irrigation potential
• complex_size_filter: Filtering by size thresholds
• complex_average: Average sizes of regions

Loading the Dataset

from datasets import load_dataset

# Load dataset
dataset = load_dataset("PeterAM4/SQuID")

# Access a sample
sample = dataset['train'][0]
image = sample['image']  # PIL Image
question = sample['question']
answer = sample['answer']  # String or numeric
type = sample['type']

# Convert answer based on type
if type in ['percentage', 'count', 'proximity_percentage', 'proximity_area',
            'building_proximity', 'building_flood_risk', 'building_fire_risk',
            'connectivity', 'size', 'total_area', 'power_calculation'] or 'complex' in type:
    answer_value = float(answer)
elif 'binary' in type:
    answer_value = int(answer)  # 0 or 1
elif type == 'fragmentation':
    answer_value = answer  # "connected" or "fragmented"

Fields

• id: Question identifier (e.g., "SQuID_0001")
• image: Satellite image path
• question: Question text with GSD notation
• answer: Ground truth answer
• type: One of 24 question types
• tier: Difficulty level (1=Basic, 2=Spatial, 3=Complex)
• gsd: Ground sampling distance in meters
• acceptable_range: [lower, upper] bounds for numeric questions (when applicable)

Evaluation

For numeric questions, check if predictions fall within the acceptable range:

import math

def evaluate(prediction, sample):
    if 'acceptable_range' in sample:
        # Numeric question - check if within human agreement range
        lower, upper = sample['acceptable_range']
        return lower <= float(prediction) <= upper
    else:
        # Non-numeric question - exact match required
        return str(prediction).lower() == str(sample['answer']).lower()

The acceptable ranges represent the natural variation in human perception for spatial measurements.

Dataset Distribution

By Tier

• Tier 1 (Basic): 710 questions (35.5%)
• Tier 2 (Spatial): 616 questions (30.8%)
• Tier 3 (Complex): 674 questions (33.7%)

Top Question Types

• complex_multi_condition: 490 questions (24.5%)
• count: 178 questions (8.9%)
• binary_comparison: 172 questions (8.6%)
• size: 166 questions (8.3%)
• percentage: 157 questions (7.8%)
• proximity_percentage: 123 questions (6.2%)
• binary_proximity: 122 questions (6.1%)
• proximity_area: 107 questions (5.3%)
• connectivity: 104 questions (5.2%)
• fragmentation: 98 questions (4.9%)

By Source

• DeepGlobe (0.5m GSD): 612 questions, 174 images - Land use classification masks
• EarthVQA (0.3m GSD): 1241 questions, 364 images - Building detection and land cover masks
• Solar Panels (0.3m GSD): 97 questions, 35 images - Solar panel segmentation masks
• NAIP (1.0m GSD): 50 questions, 14 images - Human-annotated diverse scenes

Statistics Summary

• Zero-valued answers: 102 (5.1%)
• Questions with ranges: 1577 (78.8%)
• Average questions per image: 3.4

Notes

• Questions explicitly state minimum area thresholds (e.g., "ignore patches smaller than 0.125 hectares")
• Zero-valued answers indicate absence of features (intentionally included for robustness testing)
• The benchmark tests both presence and absence of spatial features to avoid positive-only bias
• Human agreement ranges allow for natural variation in spatial perception and counting
• All measurements use metric units based on the specified GSD (Ground Sampling Distance)
• Count ranges use proportional MADc (0.19) so larger counts have wider acceptable ranges

Source Datasets & Attribution

SQuID is constructed from publicly available remote-sensing datasets. We use only images from published validation or test splits and comply with the original dataset licenses.

• DeepGlobe
  Ilke Demir et al., DeepGlobe 2018: A Challenge to Parse the Earth through Satellite Images, CVPR Workshops 2018.
  Source: https://deepglobe.org/

• EarthVQA
  Junjue Wang et al., EarthVQA: Towards Queryable Earth via Relational Reasoning-based Remote Sensing Visual Question Answering, ICCV 2023.
  Source: https://github.com/WangJunjue/EarthVQA

• Photovoltaic (Solar Panels) Dataset
  H. Jiang et al., Multi-resolution dataset for photovoltaic panel segmentation from satellite and aerial imagery, Earth System Science Data, 2021.
  Source: https://essd.copernicus.org/articles/13/5389/2021/

• NAIP Imagery
  U.S. Geological Survey, National Agriculture Imagery Program (NAIP).
  Source: https://www.usgs.gov/core-science-systems/national-geospatial-program/national-agriculture-imagery-program

All derived annotations, questions, and acceptable answer ranges introduced in SQuID are released under CC BY 4.0.

Citation

If you use this dataset, please cite:

@misc{massih2026reasoningpixellevelprecisionqvlm,
      title={Reasoning with Pixel-level Precision: QVLM Architecture and SQuID Dataset for Quantitative Geospatial Analytics}, 
      author={Peter A. Massih and Eric Cosatto},
      year={2026},
      eprint={2601.13401},
      archivePrefix={arXiv},
      primaryClass={cs.CV},
      url={https://arxiv.org/abs/2601.13401}, 
}

---