README.md

metadata

license: odbl
task_categories:
  - other
language:
  - en
tags:
  - geospatial
  - openstreetmap
  - osm
  - polygons
  - landuse
  - landcover
  - remote-sensing
  - foundation-model
size_categories:
  - 1M<n<10M

osm-polygon-selection dataset

A curated set of OpenStreetMap polygons from 310 geographic units — sovereign countries plus sub-country regions like Brazilian states, Chinese provinces, Indian zones, US states, Canadian provinces, Japanese regions, and Indonesian islands — classified by size bin (small / medium / large, area in [0.1, 100] km²) and tagged by continent (Natural Earth admin0 lookup).

Size bins:

• small — area in [0.1, 1) km² (10,000 m² to 1 km², roughly 100 m × 100 m to ~1 km × 1 km). Examples: a city block, a small park, a single farm field, a small wood lot, a residential courtyard, a parking lot, an industrial yard.
• medium — area in [1, 10) km² (1 km² to 10 km², roughly 1 km × 1 km to 3 km × 3 km). Examples: a large park, a small village/town footprint, a reservoir, a forest patch, an industrial zone, a golf course, a cemetery, a nature reserve.
• large — area in [10, 100] km² (10 km² to 100 km², roughly 3 km × 3 km to 10 km × 10 km). Examples: a large forest, a big lake, an entire town or small city, a large military training area, a national park section, a sizable agricultural region.

Polygons smaller than 0.1 km² (most individual buildings, houses, small ponds, single fields) and larger than 100 km² (whole countries, mountain ranges, big seas) are excluded by the size filter (see Filter chain below).

Status: All 310 geographic units are extracted end-to-end.

Total polygons: 16,297,690 (combined parquet: combined/all_world.parquet).

Coverage

This dataset processes one parquet per Geofabrik PBF region. Each region is bucketed in per_country/ as either a sovereign country or a sub-country region (state, province, federal district, island group, zone):

unit type	count	examples
Sovereign country	~170	france, ghana, japan, peru, australia, brazil, argentina
Sub-country region	~130	brazil-sudeste, china-beijing, japan-kanto, india-central-zone, us-texas, canada-ontario, russia-siberian-fed-district
Multi-country bundle	~10	gcc-states, ireland-and-northern-ireland, senegal-and-gambia, haiti-and-domrep, malaysia-singapore-brunei, israel-and-palestine

Total: 310 geographic units across all 6 inhabited continents plus Oceania. The 310 is not the count of sovereign states (which is ~195 per the UN); it's the count of discrete Geofabrik PBF regions we processed.

Layout

This dataset is split across five subfolders so you can pull only what you need:

folder	what's inside	typical size
per_country/	one folder per country with <country>.parquet + README.md	~28 GB total, <1 MB per small country
combined/	all_world.parquet — every polygon in one file	~14 GB
splits/	train.parquet, val.parquet, test.parquet — pre-filtered split parquets (no split column needed)	~13 GB total
sample/	sample_map.jsonl — ~18k representative polygons for quick viz	~3 MB
preview/	map_preview.png — static map thumbnail	~1 MB

Start with sample/ or preview/ for a quick look. Pull per_country/<country>/<country>.parquet for a single-country study. Use combined/all_world.parquet for cross-country work or splits/<train/val/test>.parquet for ML training with the pre-defined 80/10/10 stratified-by-country split (seed=42).

What's in this dataset

Each row is one OSM polygon (closed way or multipolygon relation) that passed our filter chain (see below). The polygon geometry itself is included in the row as WKT (or WKB if OSM_POLYGON_GEOMETRY=wkb is set when the dataset is built) so you can render, query, or reproject it directly without re-deriving from centroid+area.

column	type	description
osm_id	int64	OSM object id (int64).
osm_type	string	OSM object type, "way" or "relation" (string).
centroid_lon	float64	polygon centroid longitude (WGS84, float64).
centroid_lat	float64	polygon centroid latitude (WGS84, float64).
area_km2	float64	polygon area in km² (Web Mercator, float64).
tags	list(string)	OSM key=value tags (list of strings).
matched_tag	string	the first tag in tags that hit the whitelist (string, the reason the polygon survived).
continent	string	Natural Earth admin0 lookup of the centroid (string).
size_bin	string	"small" (0.1-1), "medium" (1-10), or "large" (10-100) km² (string).
country	string	ISO-style country name (string).
extract_status	string	"clean" (extract ran to completion) or "killed" (extract was interrupted) (string).
pbf_date	string	date of the source PBF file (string, from mtime).
geometry_wkt	string	polygon geometry as WKT (WGS84, string). Parse with shapely.wkt.loads(row.geometry_wkt).

Provenance

• Pipeline version: v0.1.0
• Git SHA: d69b105c41732c72c2162d737e3353b95bcbdfbf
• Built: 2026-07-06T23:15:51.406227
• Source: Geofabrik regional extracts (https://download.geofabrik.de/)
• Whitelist: 22,075 OSM key=value tags from osm-stats (see docs/whitelist_decisions.md in the project repo, or read the full rationale in the blog post). The whitelist is designed to filter polygons by landuse-style tags (natural, landuse, leisure, amenity, etc.) so the dataset focuses on physical land-cover / land-use features rather than buildings, addresses, or points of interest.

Geographic distribution

(Each circle is one polygon from the sample/ folder, color-coded by country. Circle size is proportional to sqrt(area_km2).)

Size-bin distribution (full dataset)

Counts every polygon in the 16,297,690-polygon dataset by size_bin, computed directly from combined/all_world.parquet via pyarrow.compute.value_counts. Percentages are exact ratios over the entire dataset, not a sample.

size_bin	count	pct
small	12,474,300	76.5%
medium	3,294,828	20.2%
large	528,562	3.2%
Total	16,297,690	100.0%

Example row

Here is one concrete row from the Liechtenstein parquet file (a natural=* polygon, fully filled-in with all 13 columns):

column	value
osm_id	1342399548
osm_type	way
centroid_lon	9.513440
centroid_lat	47.070405
area_km2	0.3202
tags	natural=grassland
matched_tag	natural=grassland
continent	Europe
size_bin	small
country	liechtenstein
extract_status	clean
pbf_date	2026-06-26
geometry_wkt	MULTIPOLYGON (((9.5109116 47.0686582, 9.5111015 47.0686316, 9.5112088 47.0685293, 9.5112785 47.06...

This row is representative: the full-dataset distribution above shows ~80% small, ~18% medium, ~2% large, and the dominant whitelist tag families (natural=*, landuse=*, leisure=*) account for the majority of matched_tag values.

Filter chain

Each polygon in this dataset has passed three filters:

1. Size filter (Stage 0): area in [0.1, 100] km². Polygons smaller than 0.1 km² or larger than 100 km² are dropped.
2. Whitelist filter (Stage 2): at least one OSM tag in the 22,075-tag whitelist. The whitelist is derived from a clustering of OSM tags across both tfidf and embeddings analyses.
3. Classify (Stage 3): continent assigned via Natural Earth admin0 shapefile, size_bin assigned by area.

Train / val / test split

Every row in every parquet (per_country/<country>/<country>.parquet and combined/all_world.parquet) carries a split column with one of three values: train, val, or test.

split	ratio	polygons
train	80%	13,037,271
val	10%	1,628,432
test	10%	1,631,987
Total	100%	16,297,690

The split is stratified by country: each country's rows are assigned to train/val/test independently using a global numpy.random.default_rng seeded once with 42 and offset per country. The exact counts per country and the seed are recorded in splits/split_manifest.json.

To load only one split (e.g. for training), filter in pyarrow:

import pyarrow.compute as pc
import pyarrow.parquet as pq
table = pq.read_table("combined/all_world.parquet")
train = table.filter(pc.equal(table["split"], "train"))

The split is deterministic and re-runnable:

uv run python scripts/make_split.py            # default seed=42, 80/10/10
uv run python scripts/make_split.py --seed 7   # different reproducible split

Per-country summary

Country	Polygons	Status
afghanistan	15,821	clean
albania	14,738	clean
algeria	32,601	clean
american-oceania	627	clean
andorra	776	clean
angola	19,197	clean
argentina	193,648	clean
armenia	7,720	clean
australia	115,145	clean
austria	133,711	clean
azerbaijan	15,826	clean
azores	2,640	clean
bahamas	2,882	clean
bangladesh	11,444	clean
belarus	223,750	clean
belgium	125,108	clean
belize	12,972	clean
benin	4,614	clean
bhutan	8,769	clean
bolivia	30,701	clean
bosnia-herzegovina	49,715	clean
botswana	5,327	clean
brazil-centro-oeste	58,137	clean
brazil-nordeste	46,565	clean
brazil-norte	46,154	clean
brazil-sudeste	59,249	clean
brazil-sul	65,807	clean
bulgaria	74,567	clean
burkina-faso	8,835	clean
burundi	4,659	clean
cambodia	4,374	clean
cameroon	71,789	clean
canada-alberta	134,835	clean
canada-british-columbia	148,372	clean
canada-manitoba	104,499	clean
canada-new-brunswick	16,313	clean
canada-newfoundland-and-labrador	91,841	clean
canada-northwest-territories	246,014	clean
canada-nova-scotia	23,043	clean
canada-ontario	182,052	clean
canada-prince-edward-island	4,579	clean
canada-quebec	251,374	clean
canada-saskatchewan	57,366	clean
canada-yukon	34,726	clean
canary-islands	2,560	clean
cape-verde	2,417	clean
central-african-republic	53,491	clean
chad	23,667	clean
chile	71,739	clean
china-anhui	17,945	clean
china-beijing	15,710	clean
china-chongqing	6,062	clean
china-fujian	13,047	clean
china-gansu	31,162	clean
china-guangdong	27,368	clean
china-guangxi	13,033	clean
china-guizhou	8,079	clean
china-hainan	4,158	clean
china-hebei	63,656	clean
china-heilongjiang	37,555	clean
china-henan	18,109	clean
china-hong-kong	1,519	clean
china-hubei	16,705	clean
china-hunan	13,182	clean
china-inner-mongolia	28,927	clean
china-jiangsu	29,605	clean
china-jiangxi	15,265	clean
china-jilin	16,150	clean
china-liaoning	12,329	clean
china-macau	90	clean
china-ningxia	6,196	clean
china-qinghai	7,691	clean
china-shaanxi	28,916	clean
china-shandong	49,948	clean
china-shanghai	5,564	clean
china-shanxi	14,141	clean
china-sichuan	22,246	clean
china-tianjin	8,836	clean
china-tibet	24,681	clean
china-xinjiang	34,254	clean
china-yunnan	16,805	clean
china-zhejiang	30,185	clean
colombia	36,412	clean
comores	395	clean
congo-brazzaville	6,643	clean
congo-democratic-republic	85,106	clean
costa-rica	7,476	clean
croatia	47,140	clean
cuba	23,386	clean
cyprus	4,846	clean
czech-republic	271,062	clean
denmark	175,795	clean
djibouti	576	clean
east-timor	1,553	clean
ecuador	16,139	clean
egypt	24,623	clean
el-salvador	3,517	clean
equatorial-guinea	1,004	clean
eritrea	3,278	clean
estonia	47,160	clean
ethiopia	29,663	clean
faroe-islands	1,278	clean
fiji	3,093	clean
finland	427,870	clean
france	492,538	clean
gabon	3,843	clean
gcc-states	59,856	clean
georgia	21,198	clean
germany	1,131,888	clean
ghana	11,445	clean
greece	45,142	clean
greenland	15,132	clean
guernsey-jersey	670	clean
guinea	12,311	clean
guinea-bissau	2,109	clean
guyana	2,192	clean
haiti-and-domrep	4,604	clean
honduras	6,160	clean
hungary	77,569	clean
iceland	47,896	clean
india-central-zone	51,347	clean
india-eastern-zone	21,480	clean
india-north-eastern-zone	12,160	clean
india-northern-zone	50,862	clean
india-southern-zone	66,323	clean
india-western-zone	39,173	clean
indonesia-java	10,582	clean
indonesia-kalimantan	7,663	clean
indonesia-maluku	3,800	clean
indonesia-nusa-tenggara	7,778	clean
indonesia-papua	4,208	clean
indonesia-sulawesi	5,805	clean
indonesia-sumatra	10,121	clean
iran	57,028	clean
iraq	26,861	clean
ireland-and-northern-ireland	159,879	clean
isle-of-man	2,648	clean
israel-and-palestine	13,681	clean
italy	276,991	clean
ivory-coast	14,273	clean
jamaica	1,865	clean
japan-chubu	14,376	clean
japan-chugoku	10,128	clean
japan-hokkaido	18,638	clean
japan-kansai	9,902	clean
japan-kanto	16,024	clean
japan-kyushu	23,846	clean
japan-shikoku	4,686	clean
japan-tohoku	18,838	clean
jordan	4,078	clean
kazakhstan	78,493	clean
kenya	16,916	clean
kiribati	584	clean
kosovo	5,377	clean
kyrgyzstan	17,110	clean
laos	6,380	clean
latvia	47,133	clean
lebanon	4,289	clean
lesotho	19,246	clean
liberia	2,342	clean
libya	13,046	clean
liechtenstein	565	clean
lithuania	76,550	clean
luxembourg	11,460	clean
macedonia	9,330	clean
madagascar	22,664	clean
malawi	5,337	clean
malaysia-singapore-brunei	20,439	clean
maldives	2,358	clean
mali	40,392	clean
malta	620	clean
marshall-islands	628	clean
mauritania	9,040	clean
mauritius	1,863	clean
mayotte	606	clean
mexico	68,967	clean
micronesia	755	clean
moldova	35,690	clean
monaco	2	clean
mongolia	10,941	clean
montenegro	11,785	clean
morocco	42,623	clean
mozambique	11,101	clean
myanmar	32,570	clean
namibia	9,316	clean
nepal	68,869	clean
netherlands	207,459	clean
new-caledonia	2,141	clean
new-zealand	127,834	clean
nicaragua	10,633	clean
niger	14,606	clean
nigeria	33,059	clean
north-korea	19,295	clean
norway	413,801	clean
pakistan	36,200	clean
panama	6,663	clean
papua-new-guinea	9,006	clean
paraguay	29,619	clean
peru	26,038	clean
philippines	36,555	clean
poland	637,908	clean
polynesie-francaise	2,195	clean
portugal	66,287	clean
romania	115,401	clean
russia-central-fed-district	315,956	clean
russia-crimean-fed-district	20,645	clean
russia-far-eastern-fed-district	192,337	clean
russia-kaliningrad	12,432	clean
russia-north-caucasus-fed-district	52,352	clean
russia-northwestern-fed-district	381,413	clean
russia-siberian-fed-district	291,830	clean
russia-south-fed-district	166,644	clean
russia-ural-fed-district	184,300	clean
russia-volga-fed-district	237,169	clean
rwanda	3,976	clean
saint-helena-ascension-and-tristan-da-cunha	174	clean
samoa	369	clean
sao-tome-and-principe	190	clean
senegal-and-gambia	20,479	clean
serbia	47,189	clean
seychelles	264	clean
sierra-leone	3,366	clean
slovakia	54,888	clean
slovenia	41,526	clean
solomon-islands	4,575	clean
somalia	48,453	clean
south-africa	120,252	clean
south-korea	39,039	clean
south-sudan	17,091	clean
spain	240,230	clean
sri-lanka	13,748	clean
sudan	38,043	clean
suriname	9,410	clean
swaziland	4,113	clean
sweden	397,661	clean
switzerland	61,156	clean
syria	21,603	clean
taiwan	14,124	clean
tajikistan	24,831	clean
tanzania	16,046	clean
thailand	32,789	clean
togo	3,408	clean
tonga	916	clean
tunisia	8,498	clean
turkey	113,609	clean
turkmenistan	6,353	clean
uganda	10,449	clean
ukraine	645,578	clean
united-kingdom	205,002	clean
uruguay	10,051	clean
us-alabama	28,004	clean
us-alaska	62,521	clean
us-arizona	38,727	clean
us-arkansas	18,513	clean
us-california	51,998	clean
us-colorado	61,550	clean
us-connecticut	12,802	clean
us-delaware	7,255	clean
us-district-of-columbia	564	clean
us-florida	229,464	clean
us-georgia	29,945	clean
us-hawaii	2,619	clean
us-idaho	17,217	clean
us-illinois	79,460	clean
us-indiana	42,034	clean
us-iowa	30,362	clean
us-kansas	56,789	clean
us-kentucky	12,179	clean
us-louisiana	17,618	clean
us-maine	17,529	clean
us-maryland	24,684	clean
us-massachusetts	22,134	clean
us-michigan	73,209	clean
us-minnesota	72,441	clean
us-mississippi	7,704	clean
us-missouri	48,147	clean
us-montana	18,606	clean
us-nebraska	31,375	clean
us-nevada	14,207	clean
us-new-hampshire	11,186	clean
us-new-jersey	28,145	clean
us-new-mexico	21,070	clean
us-new-york	42,871	clean
us-north-carolina	36,095	clean
us-north-dakota	14,275	clean
us-ohio	117,344	clean
us-oklahoma	13,752	clean
us-oregon	45,342	clean
us-pennsylvania	44,038	clean
us-puerto-rico	2,318	clean
us-rhode-island	6,360	clean
us-south-carolina	20,613	clean
us-south-dakota	11,975	clean
us-tennessee	22,002	clean
us-texas	82,221	clean
us-utah	22,027	clean
us-vermont	7,452	clean
us-virgin-islands	193	clean
us-virginia	33,844	clean
us-washington	51,195	clean
us-west-virginia	25,081	clean
us-wisconsin	84,094	clean
us-wyoming	14,739	clean
uzbekistan	35,052	clean
vanuatu	663	clean
venezuela	14,723	clean
vietnam	21,025	clean
yemen	7,008	clean
zambia	15,684	clean
zimbabwe	7,545	clean
Total	16,297,690

Back to the dataset root