Where are the cameras?
Every license-plate reader OpenStreetMap carries inside one metro window — mapped by volunteers who walked or drove past the poles. The window is deliberately wider than Alpharetta: it holds Roswell, Johns Creek, and Milton too, because any box drawn over one of these cities contains three others.
Interactive map of mapped plate readers
● OpenStreetMap only · ● OSM and Alpharetta PD’s released list describe it — the position shown is the agency’s · ● released list only — OSM has never mapped it
Browse the sites as a list
Text alternative: 726 mapped plate-reader cameras stand inside the metro window bounded by 33.96°N to 34.18°N, 84.45°W to 84.1°W. 134 fall inside Alpharetta’s city boundary; the remaining 592 are in Roswell, Johns Creek, Milton and the surrounding edges. They occupy 662 distinct sites, 56 of which carry more than one camera. By brand: Flock Safety 590, Genetec 8, Axon Enterprise 2, Axis Communications 1, Leonardo 1, Motorola Solutions 1, and 123 with no brand tag. Every count on this map is stated in the panels below.
Is there one near me?
That is the question most people arrive with, and a field of 662 dots does not answer it. Type a street into the box above the map — find your street — and the map goes there and marks what stands within 250 feet of it. Every named road in the window is in the list, all 11,431 of them, across 19 cities and counties.
Have at least one mapped camera within 250 feet. The other 10,675 have none — which is most of them, and is the answer most people get. A camera covers a road, not a neighborhood: these are arterials, interchanges and the entrances to subdivisions, not the streets inside them.
For a street with no camera on it, that is the median distance to the closest one — about 0.4 miles, or the length of a short errand. Half of these streets are closer than that. The lookup says which it is for yours, in feet, either way.
The index is one 693 KB file from this origin, fetched once and searched inside your browser. No geocoder, no autocomplete service, no request per keystroke. A site about passive location trails should not make one out of its own search box, and this one does not: the street you look up is not sent anywhere, because there is nowhere for it to be sent.
Near means within 250 feet, and that is all it means. It is not a claim about which road a camera is pointed at: a camera on a corner is within 250 feet of both roads and is listed under both, because deciding which one it watches is a guess this data cannot support. One name can be two streets — this window holds four cities and several share street names — so each is listed with the municipality it falls in. And the same limit governs here as everywhere else on this page: a street is listed because a volunteer mapped it, and a camera is near it because a volunteer mapped that too. An empty result is an absence of evidence, not evidence of absence. Source: streets.json, built from the same OpenStreetMap extract the roads under the map are drawn from.
How to read this map
“How many cameras in this city” has four defensible answers that differ by more than half: a rectangle over this window returns 285, a circle 138, nearest-place 150, and the real boundary 134. Only the last is correct — and it is still a floor, because drive-by mapping misses private streets.
A camera in Alpharetta may be run by the county, the state DOT, a school district, an HOA, or a store. The map names an operator for 22 of the 134, and just 2 say Alpharetta Police — against a department running about 108 devices. The most common operator value nationally is the manufacturer.
A pole with three lenses is three cameras in the data. Dots here merge cameras within 30 metres into one site — 726 cameras are 662 sites, and 56 sites hold more than one. Plotting one pin per camera would stack them invisibly.
One more honest limit: the ALPR tag is a mapper’s judgment from the sidewalk. Half of Flock’s product line does not read plates, and a video or audio unit in a Flock housing could be counted here with nothing in the data to find it.
Whose hardware is on the poles
A brand tag is what a mapper could read off the housing, so “no brand tag” is a gap in the mapping, not a generic camera. Nationally Flock Safety is 82% of the 130,550 plate readers mapped across the US and Canada. Source: alpharetta-metro-cameras.geojson.
The same map, at national scale
Alpharetta’s 134 is a number without a scale until it stands next to the rest of the country. The same volunteer survey that found those 134 has recorded 128,855 plate readers nationally, 106,377 of them Flock, against 2,685 agencies the EFF’s Atlas of Surveillance records as Flock customers. Switch the map above to United States to see how unevenly that falls. The join returns 49 states plus DC and Puerto Rico; Alaska appears in neither source and has no row at all.
Georgia carries 8,352 mapped Flock cameras across 150 agencies — 55.7 per agency, against a national 39.6. Alpharetta’s 134 inside the city boundary is 1.6% of the state’s mapped total, in a city holding well under one percent of its people.
The top five states hold 48% of every Flock camera on the map. That is partly deployment and partly survey effort, and the data cannot separate them — which is the honest reading of every choropleth built from volunteer mapping.
The two sources measure different things: OpenStreetMap records cameras somebody stood in front of, the Atlas records agencies somebody documented buying. Where the second runs far ahead of the first, the mapping is behind — not the deployment. These five have the fewest mapped cameras per agency known to run Flock, against a national 39.6.
The full table 51 rows · every state the join returns
Per-state join of OSM camera counts against EFF Atlas agency records. States assigned by point-in-polygon on a simplified boundary file; points within ~1km of a border may be misassigned. Ranked by mapped Flock cameras per known Flock agency, ascending. Low values indicate states where deployment is documented but mapping lags — the highest-value targets for field survey. Confounder: a state can rank low simply because its agencies run small deployments. Neither source is a register of deployed devices, and this site does not present them as one. Source: states.json and coverage_gap.json, joined here from 128,855 camera records and 15,181 Atlas deployment records held in this project’s own store — nothing on this page is fetched from either upstream at read time.