OpenStreetMap · extracted · deduplicated · classified

From a continent to a single intersection.

One clean record per real signalized intersection — across 225 cities in the US, 271 in Europe and 43 in Canada — classified and ready for simulation, V2X and HD maps.

Vehicular Pedestrian Both HAWK Level crossing
539 cities · US · Europe · Canada
170,000+
US intersections
539
cities · US · EU · Canada
41
European countries
4,856
LA signals · matches LADOT (~4,850)

The problem

Raw OpenStreetMap is a cloud of poles. I turn it into intersections.

A single junction shows up in OpenStreetMap as several signal poles, tagged inconsistently, with nothing tying them together. MobilityMetric resolves all of that into one clean, classified record per real intersection — each with a nearest street and a confidence score.

node 653656 highway=traffic_signals node 653688 highway=traffic_signals node 653701 crossing=traffic_signals ⋮ several poles, one junction ────────────▶ one clean record { "intersection_id": "US-CA-SOCAL-LOS-9q5ctr1p", "signal_type": "vehicular_junction", "approach_count": 4, "nearest_street": "Wilshire Blvd" }

Three live datasets, one pipeline

Pick your coverage.

US · 50 states + DC

US Traffic Signals

225 cities

~170,000 signalized intersections, classified by type and matched to nearest streets — with counts that track published city figures.

Europe · 41 countries

Europe Traffic Signals

271 cities

One point per real intersection, with nearest street and a confidence score, kept within actual city limits.

Canada · 13 provinces

Canada Traffic Signals

43 cities

Signalized intersections across every province and territory, classified by type and combined with official open data from Toronto, Montréal, Calgary and Winnipeg.

Europe · 12 countries

EU LEZ Intelligence

94 zones

Low Emission Zone polygons with rules, areas and route-compliance evaluation — another dataset I'm building out in this project.

Proof

It tracks the official city figures.

Vehicular signalized-intersection counts land within a few percent of what city transportation departments publish.

CityOfficial DOTThis datasetDifference
Los Angeles~4,850 (LADOT)4,8560%
New York~13,770 (NYCDOT)13,198−4%
Chicago~2,900 (CDOT)2,988+3%

Feel the data

Search any city. See up to 500 real intersections.

Pick a region and a city, run a sample, and inspect each intersection on a live map — signal type, nearest street and confidence — then download it as GeoJSON or CSV.

live · US · Europe · Canada

About this project

A data-engineering project, start to finish.

MobilityMetric is a personal project: I built the pipeline that turns open map data into one clean, classified record per real intersection across the US, Europe and Canada. The live map runs a capped sample of up to 500 intersections per city.

Open the live map