GeoPathfinder

A Pathfinder-style long-range spatial reasoning benchmark on real satellite imagery. Each sample is a 256×256 Landsat patch (native 30 m resolution, a 7.7×7.7 km footprint) with two red dots on land; the task is to predict whether the dots are connected by land (connect = 1) or separated by water (disconnect = 0). Labels derive from Overture Maps water polygons rasterized onto the exact pixel grid of each patch and cross-checked against Landsat's own QA water flag.

• 5,553 zero-overlap patches across 42 river basins on six continents (no two patch centers within one patch width)
• 5,273 benchmark patches → 10,546 images, exactly class-balanced (each patch yields one connect and one disconnect sample, distance-matched within 25% so dot separation carries no label signal)
• Adversarial dot placement: each released pair is the most deceptive of a large candidate pool (96 for train/val, 256 for test), deterministic given the masks — randomly placed dots admit a sight-line shortcut
• Fixed 80/10/10 patch-level splits (paired images never straddle a split); snow scenes and unlabelable patches excluded. A basin-disjoint protocol (splits_basin.json, repo root) holds out four basins on four continents for geographic-generalization tests
• eval_variants/ ships a fixed easy/medium/hard placement suite (1,051 images per tier): report per-tier accuracy and the macro average (placement-general score) alongside test accuracy
• Code, construction pipeline, baselines: https://github.com/isaaccorley/geo-long-range-arena

Download

The full dataset ships as a single archive that extracts to a geopathfinder/ folder:

hf download isaaccorley/GeoPathfinder geopathfinder.tar.gz --repo-type dataset --local-dir .
tar -xzf geopathfinder.tar.gz

metadata.parquet and water_polygons.parquet are also available uncompressed at the repo root for querying without downloading the archive.

Files (inside geopathfinder.tar.gz)

pathfinder/connect/     patch_XXXXX.jpg    benchmark image, label = 1
pathfinder/disconnect/  patch_XXXXX.jpg    benchmark image, label = 0
images/                 patch_XXXXX.jpg    true color without dots (pristine)
raw/                    patch_XXXXX.tif    uint16 RGBN surface-reflectance DNs,
                                           scale/offset in band metadata, scene
                                           provenance in file tags (ZSTD)
masks/                  patch_XXXXX.png    0=land 1=water 2=invalid 3=suspect
overlays/               patch_XXXXX.jpg    water mask blended over true color (QC)
metadata.parquet                           one row per patch: split, region, scene
                                           id, bounds, CRS/transform, QC stats,
                                           dot pixel + lon/lat coordinates
water_polygons.parquet                     281k Overture water polygons clipped to
                                           patch bounds (GeoParquet, WGS84), keyed
                                           by sample_id; geometry_patch_crs holds
                                           the same shapes as WKB in each patch's
                                           UTM CRS (patch_crs column)
splits.json                                train/val/test patch ids
snowy_ids.json                             excluded snow-scene ids
plan.json                                  build configuration snapshot

Rasterizing a patch's rows from water_polygons.parquet onto its transform/crs (in metadata.parquet) reproduces its stored mask (IoU ≥ 0.999).

Quick start

import pandas as pd

meta = pd.read_parquet("metadata.parquet")
train = meta[meta.split == "train"]
# each labeled patch contributes two images:
#   pathfinder/connect/{sample_id}.jpg    label 1
#   pathfinder/disconnect/{sample_id}.jpg label 0

Raw reflectance (for geospatial foundation models — dots are not burned into these; use the dot coordinates from metadata.parquet):

import rasterio

with rasterio.open("raw/patch_00122.tif") as src:
    reflectance = src.read().astype("float32") * src.scales[0] + src.offsets[0]

Baselines

Two numbers per model: accuracy on the released test split, and the placement-general score (macro accuracy over the easy/medium/hard suite, seed-42 checkpoints). Accuracies are mean ± std over three seeds.

Reference pipelines and probes:

approach	test acc	macro
Water index (blue-red on RGB), no learning	52.8%	54.1
U-Net masks -> connected components	57.5%	66.6
Water index (NDWI on green/NIR), no learning	69.9%	76.2
Best frozen linear probe (13 pretrainings)	62.7%	—

Architectures (fine-tuned on the released placements):

model	test acc	macro
OverLoCK-B	88.5 ± 0.5%	62.4
FocalNet-B LRF	87.7 ± 0.4%	60.9
MaxViT-B	87.5 ± 0.9%	63.0
ConvNeXt-B	87.4 ± 1.0%	61.3
UniRepLKNet-T	87.0 ± 0.4%	61.4
FocalNet-B SRF	86.9 ± 0.8%	61.0
DaViT-B (2/3 seeds converge)	86.3 ± 0.6%	60.8
MambaOut-B	86.2 ± 0.4%	60.3
MambaVision-B	85.9 ± 0.1%	61.5
ResNet-50	85.3 ± 0.6%	60.2
Swin-B	83.7 ± 1.3%	56.1
RepLKNet-31B	82.9 ± 2.9%	53.9
ViT-B/16	79.3 ± 1.0%	56.8
ResNet-18 (scratch)	77.4 ± 0.9%	53.6

Every model trained on the released placements collapses on the macro score (models learn "the obvious answer is wrong" and fail easy placements); retraining with dots resampled every epoch recovers up to 85.1 ± 0.3 macro (MaxViT) while the hard tier stays around 75% — the benchmark is far from solved. Marker shape and color barely matter (±2 points). Full protocol and training recipes: https://github.com/isaaccorley/geo-long-range-arena

Provenance and licensing

• Imagery: Landsat Collection 2 Level-2 (USGS, public domain), accessed via the Microsoft Planetary Computer. Scene ids, WRS path/row, and acquisition times are embedded in each GeoTIFF and in metadata.parquet.
• Water geometry: Overture Maps base/water (release 2026-06-17.0), which includes OSM-derived data © OpenStreetMap contributors. The water_polygons.parquet and masks/ layers are therefore available under ODbL; annotations and imagery composites are CC-BY-4.0.

Citation

@misc{corley2026geopathfinder,
  title  = {GeoPathfinder: A Long-Range Spatial Reasoning Benchmark on Real Satellite Imagery},
  author = {Corley, Isaac},
  year   = {2026},
  url    = {https://huggingface.co/datasets/isaaccorley/GeoPathfinder}
}