ray0rf1re/urbex

---
license: mit
language:
- en
tags:
- urbex
- urban-exploration
- abandoned-places
- geospatial
- kml
- google-earth
- safety
size_categories:
- 100K<n<1M
---

# Urbex — Global Abandoned Locations Dataset (100k+)

> **Safety and Legal Notice:** Urban exploration may constitute trespass and
> can be extremely dangerous. Always research local laws, never explore alone,
> and use appropriate PPE. This dataset is for research and documentation
> purposes only. The maintainers do not encourage illegal entry.

## Overview

A comprehensive dataset of **101,000 abandoned and derelict locations**
worldwide, combining real places sourced from OpenStreetMap with procedurally
generated synthetic entries modelled on real-world distributions.

Each record includes hazard assessments, access difficulty ratings, recommended
PPE and tools, and a safety score.

## Files

| File | Format | Description |
|------|--------|-------------|
| `urbex_full.kml` | KML (Google Earth) | All 101,000 locations as placemarks, colour-coded by safety |
| `data_NNNN.parquet` | Parquet | Chunked tabular data (10,000 rows each) |
| `data.csv` | CSV | Full flat export |

### Opening the KML

Download `urbex_full.kml` and open in:
- **Google Earth Pro** (File > Open)
- **Google Earth Web** (drag and drop)
- **QGIS** (Layer > Add Vector Layer)
- **ArcGIS** (Add Data)

Placemark colours: Green = Safety 20-25 | Yellow = 15-19 | Orange = 8-14 | Red = 0-7

## Schema

| Column | Type | Description |
|--------|------|-------------|
| `id` | string | Unique ID (`osm_` prefix = real, `syn_` = synthetic) |
| `name` | string | Location name |
| `latitude` | float | WGS84 latitude |
| `longitude` | float | WGS84 longitude |
| `country` | string | Country |
| `city` | string | Nearest city or town |
| `location_type` | string | Category (factory, hospital, asylum, ...) |
| `type_label` | string | Human-readable type label |
| `year_abandoned` | int | Estimated year of abandonment |
| `description` | string | Narrative description |
| `hazards` | JSON list | Hazard keys present at this site |
| `hazard_descriptions` | string | Full hazard descriptions with safety guidance |
| `access_difficulty` | int | 1 (walk-in) to 10 (practically inaccessible) |
| `access_notes` | string | Perimeter / barrier description |
| `recommended_tools` | JSON list | Recommended PPE and tool keys |
| `tool_descriptions` | string | Full tool descriptions |
| `safety_rating` | int | 0 (extremely dangerous) to 25 (relatively safe) |
| `source` | string | `openstreetmap` or `synthetic` |
| `osm_tags` | JSON string | Raw OSM tags for real records |
| `verified_date` | string | Year data was compiled |

## Statistics

| Metric | Value |
|--------|-------|
| Total records | 101,000 |
| Unique countries | 42 |
| Avg safety rating | 17.1/25 |
| Avg access difficulty | 2.2/10 |

**Top location types:**
| Type | Count |
|------|-------|
| factory | 12,118 |
| hospital | 8,516 |
| school | 7,164 |
| warehouse | 7,086 |
| church | 6,422 |
| asylum | 5,599 |
| farmhouse | 4,919 |
| hotel | 4,869 |
| mine | 4,489 |
| station | 4,154 |

**Top hazards:**
| Hazard | Count |
|--------|-------|
| structural_collapse | 54,811 |
| broken_glass | 51,412 |
| mold | 45,297 |
| lead_paint | 43,521 |
| asbestos | 40,779 |
| chemical_contamination | 23,395 |
| animal_infestation | 20,497 |
| sharp_metal | 19,185 |
| pcbs | 14,535 |
| water_damage | 12,738 |

## Hazard Key

| Key | Description |
|-----|-------------|
| `asbestos` | Asbestos-containing materials |
| `mold` | Black mold / mycotoxin contamination |
| `water_damage` | Flooding, standing water, bacterial risk |
| `structural_collapse` | Compromised load-bearing elements |
| `chemical_contamination` | Industrial chemical residue |
| `lead_paint` | Pre-1978 lead-based paint |
| `electrical_hazard` | Live or exposed wiring |
| `radiation` | Ionising radiation (medical or industrial) |
| `pcbs` | Polychlorinated biphenyls |
| `mercury` | Elemental mercury contamination |
| `silica_dust` | Respirable crystalline silica |
| `animal_infestation` | Rodents, bats, birds (disease risk) |
| `fire_damage` | Fire-weakened structure and toxic char |
| `underground_voids` | Hidden sub-floor voids or mine shafts |
| `confined_spaces` | Oxygen-deficient or toxic-gas risk spaces |
| `security_patrol` | Active security on site |
| `alarm_system` | Active intruder alarm |

## Safety Rating Scale

| Score | Category |
|-------|----------|
| 21-25 | Relatively Safe |
| 16-20 | Moderate Risk |
| 11-15 | Dangerous |
| 6-10 | Very Dangerous |
| 0-5 | Extreme Danger — specialist access only |

## Access Difficulty Scale

| Rating | Meaning |
|--------|---------|
| 1-2 | Open access, no barrier |
| 3-4 | Minor barrier (low fence, gap in hoarding) |
| 5-6 | Significant barrier (chain-link, locked gate) |
| 7-8 | Serious barrier (razor wire, wall, CCTV) |
| 9-10 | Extreme (security guards, dogs, 24/7 monitoring) |

## Data Sources

- Real records: OpenStreetMap contributors via Overpass API
(licensed ODbL 1.0: https://opendatacommons.org/licenses/odbl/)
- Synthetic records: Procedurally generated using real-world city coordinates
and statistically modelled distributions of abandonment types, years and hazards.

## Related

- Real-only dataset: [ray0rf1re/urbex-sml](https://huggingface.co/datasets/ray0rf1re/urbex-sml)

## Citation

```bibtex
@dataset{urbex2024,
title = {Urbex — Global Abandoned Locations Dataset},
author = {ray0rf1re},
year = {2024},
url = {https://huggingface.co/datasets/ray0rf1re/urbex},
license = {MIT}
}
```