Perch Scout
← Back to the map

About

Perch Scout helps you find Flock surveillance cameras that aren't on OpenStreetMap yet, so you can go check them in person.

How to use this site

  1. Find an unmapped cameraUse the map to find one near you that isn't on OpenStreetMap yet.
  2. Go see it in personCheck if it's really there. Only add what you saw yourself.
  3. Add it to OpenStreetMapDownload the DeFlock app, or add it directly on OpenStreetMap.

About the data

The map compares Flock's own records of its devices with the cameras people have already put on OpenStreetMap. The gaps are where to go look.

Plate reader cameras

Not yet mapped—
Might already be mapped—
Planned in 2025—
On OpenStreetMap already—

Gunshot detectors

Not yet mapped—
Might already be mapped—
Planned in 2025—
On OpenStreetMap already—

In the US. "Planned" devices weren't installed yet in December 2025, so they aren't counted in the first two columns.

Where it comes from

What the colors mean

Not yet mapped (go check it!)
In Flock's records. Nothing on OpenStreetMap nearby could be it.
Might already be mapped
A mapped camera nearby might or might not be this one. It faces another way, it's a bit far, or another device next to this one fits it better.
Planned in 2025
Not installed yet when the list was made. It may be up now.
On OpenStreetMap and Flock's list
Someone has already mapped it, and it matches a device in Flock's records. Tap it to see what Flock's records say.
On OpenStreetMap
Mapped, but no device in Flock's records matches it. It may be newer than 2025, or from another maker.
Decommissioned
Taken out of service by 2025. It might still be on the pole.

How we compare them

We pair each device in Flock's records with at most one OpenStreetMap camera, and each OpenStreetMap camera with at most one device. Closer pairs, and pairs that face the same way, are paired first. A device counts as mapped when its pair is very likely the same camera.

The locations in Flock's records are often off, sometimes by 100 m or more. OpenStreetMap is usually closer, since someone stood there and mapped it. So a mapped camera 100 m away that faces the same way, with no other device next to it, usually counts.

Some devices are listed more than once in Flock's records. We show those once. We leave out test units piled up at Flock's own sites, and portable units, which move.

How the matching works, step by step

1. Clean up Flock's records. When 10 or more devices sit on the exact same spot, they're factory or test units, not cameras on poles, so we leave them out. We also leave out portable Falcon Flex units and plate reader trailers. They move, and their records often give an office address. Devices listed twice become one: copies under a property manager's name, a planned and an installed record for the same spot, and "Original" records left behind after a camera moved.

2. Score every possible pair. For each device, we look at every OpenStreetMap camera within 400 m and estimate how likely it is to be the same camera. Distance sets the starting point: about 99% for pairs under 10 m, 70% around 60 m, 50% around 130 m, and 30% around 250 m. These numbers come from the data itself. At each distance, we checked how often pairs face the same way more than chance would explain.

3. Adjust for direction. Facing the same way makes a match about five times more likely. Facing a different way makes it less likely, but not by much. Directions are often wrong on both sides, and nobody edits OpenStreetMap just to turn a camera. And if nothing else could be that mapped camera, and the mapped cameras nearby at least match Flock's count, we ignore direction altogether: whoever mapped that corner wouldn't have missed this camera. A mapped camera facing exactly north counts less, because that's often a placeholder. When Flock's records have no exact angle, we use the direction in the label ("WB" means westbound), which also counts less.

4. Pair them up, best first. Each device and each OpenStreetMap camera is used once, so two devices can't both match one mapped camera. Mappers often put every camera on a pole on one point, with one direction each, so we count each direction as a camera.

5. Sort them. At least 80% likely: already mapped, and hidden from the map. Between 35% and 80%: might already be mapped. The details say why: it's a bit far, it faces a different way, or another device next to it fits better. Anything else is not yet mapped, even with a mapped camera close by. That one is probably a different camera.

We'll retune these numbers as people check leads in person.

What this can't tell you

What we collect

Nothing. See Privacy for the other services your browser talks to.

FAQ

How do I add a camera I found?

Use the DeFlock app or the OpenStreetMap editor. You'll need a free OpenStreetMap account. Put the pin where you saw the camera and set the direction you saw it facing. Don't copy the location or direction from Flock's records.

I checked and the camera isn't there. Can I remove it from this site?

Not yet. Reporting a missing camera is a planned feature. For now, just don't add it to OpenStreetMap. The dot stays on this map because it comes from the December 2025 snapshot of Flock's records, which doesn't change.

If a camera that's already on OpenStreetMap is gone, you can remove it in OpenStreetMap.

How do you know which way a camera faces?

Many devices in Flock's records come with an exact angle. When one doesn't, we read Flock's label: "WB" means it watches westbound traffic, so it faces west. That guess is right about 85% of the time. On the map, guessed directions get a wider cone.