Where was this photo taken?
There are five routes to the answer, from a check that takes seconds to handwork that fills an afternoon. Which one you need comes down to a single question: is the metadata still intact?
Always start here
Before reaching for anything complicated: open the file and see whether the location is still in it. For a photo straight off a camera or phone there is a real chance the coordinates are simply there and you are done in ten seconds. For a photo that reached you through WhatsApp, Instagram or a screenshot, that data is almost certainly gone and you can move straight on to the next method.
That distinction shapes everything that follows, so it is worth establishing first rather than guessing.
The five methods
Read the EXIF data
Cameras and phones write a block of metadata into the JPEG at the moment of capture. If it contains a GPS section you have the coordinates immediately. This is always the first step: it takes seconds and answers the question outright when the data is there.
Limitation: In practice the data has usually already been removed. Every messaging app and social platform strips metadata on send.
Check the EXIF data in your photo →Ask the source
The simplest method is the one most often skipped. If the photo came from a colleague, a customer or a public account, asking where it is beats any technical trick on both speed and reliability. Ask for the original file while you are at it, since that often still carries the metadata the forwarded copy lost.
Limitation: No use with anonymous sources, and the answer cannot be independently verified.
Reverse image search
Google Images, Yandex, TinEye and Bing can find an image on other pages. If the same photo has been published before with a caption, you have your answer. Yandex is notably strong at recognising buildings and landscapes even when the exact image has never been indexed.
Limitation: Useless for photos that have never been online, and the captions you find can themselves be wrong.
Compare GeoPin with Google Lens →Read the clues in the image
This is the classic OSINT approach. The Netherlands is unusually readable: red asphalt on cycle paths, blue cycle signage, street name plates that differ by municipality, brick paving patterns, facade styles that date a street to a building period, the angle of the sun and even the planting in the verge. Combine two or three of those signals and the search area shrinks fast enough to check by hand in a street imagery viewer.
Limitation: Takes time and demands regional knowledge. An anonymous new-build estate or an enclosed interior gives you very little to work with.
Read the practical OSINT guide →Let AI match the image
Instead of reading clues yourself, you have a model compare the photo against an index of street imagery. The model ignores metadata and looks at what is actually visible, searching for the reference images that most resemble yours. That scales to millions of images and does in seconds what manual searching takes hours to do.
Limitation: Requires the location to be in the reference index. Interiors, aerial shots and areas without street imagery coverage fall outside it.
Try it with your own photo →How Dutch photos give themselves away
Once the metadata is gone the photo itself has to do the work. The Netherlands is a rewarding place for that: the public realm is heavily standardised and at the same time regionally distinct, so a handful of details can rule out a great deal.
Cycle paths
Red asphalt with white lane markings is close to uniquely Dutch. The profile, the separation from the carriageway and the colour of the paving vary by municipality and by the period the street was laid out.
Signage
The blue ANWB cycle signs, the white hectometre posts and the shape of street name plates differ by region and by the body that maintains them. A single sign can narrow the search to one province.
Facades and building period
Amsterdam canal houses, Rotterdam post-war reconstruction, Brabant nineteen-thirties housing and Vinex new-build each have their own silhouette, window rhythm and brick colour.
Street furniture
Lamp posts, bins, bike stands and benches are procured per municipality. The same post usually stands throughout that municipality and nowhere else.
Water and dykes
The relationship between road, verge, ditch and dyke often gives away the region. A quay in Zeeland looks materially different from a polder embankment in the Green Heart.
Sun and shadow
The direction and length of shadows give the time of day and the season, which helps rule candidates out and lets you test a claim about when the photo was taken.
A worked example
Say someone forwards you a photo of a street of terraced houses, a red cycle path and a church tower in the background. The EXIF is empty, because it arrived over WhatsApp.
Manually you would reason like this: the red cycle path with paved separation points to a municipality that redeveloped after roughly 1990. The facades with bay windows and a steep roof fit nineteen-thirties housing. The church tower is a concrete anchor point, but there are hundreds in the Netherlands, so you now have comparing to do. Budget an hour or more, and accept that you may not find it.
Image matching skips that reasoning. The model turns the photo into a numeric description of what it shows and searches for the nearest descriptions in an index of Dutch street imagery. What comes back are the reference images that resemble it most, with their coordinates. So you get not just an answer but the evidence beside it, and can judge the match yourself.
What GeoPin does
GeoPin is that last method, built specifically for the Netherlands. You upload an outdoor photo and get the most likely location back within seconds, with coordinates, address and postcode, plus the street imagery the match rests on. No metadata is needed and none is used.
Coverage is nationwide, with the highest image density in the major cities. To see what happens under the bonnet, read how GeoPin works. If you are building something yourself, the API documentation is ready for you.
Frequently asked questions
Can I find where a photo was taken without GPS data?
Yes. GPS data is the fastest route, not the only one. Once the metadata is gone the content of the photo remains, and it almost always holds enough clues to find the place. That can be done manually with OSINT techniques or automatically with image matching against a street imagery index.
Why does WhatsApp remove the location from my photo?
For privacy. Without that step every forwarded photo could give away the sender's home address, workplace or holiday destination to everyone in the group. WhatsApp, Signal, Instagram, Facebook and X all do this, and compress the image while they are at it.
How precisely can the location be determined?
It depends on the method and on the photo. GPS data from a phone outdoors usually gets within 5 to 10 metres. Image matching against street imagery gets within a few metres on a distinctive street, and returns an area of tens to hundreds of metres in harder cases. Manual OSINT work can reach metre precision but costs considerably more time.
Does this work for old or scanned photos?
Partly. A scanned print never had EXIF location, so that route is out. Image matching does work, as long as the scene still resembles the current street view closely enough. For photos predating major redevelopment or demolition it gets difficult quickly.
Can I locate a photo taken abroad?
Not with GeoPin. Our reference index covers the Netherlands. For photos from elsewhere the manual OSINT methods and reverse image search are the routes to take.
Is it legal to work out where a photo was taken?
Determining the location of an image is not in itself prohibited. What you then do with that information does fall under the GDPR among other things. Tracing a person's home address or assembling a profile touches on personal data and needs a lawful basis. See our acceptable use policy for where we draw the line.
Try it with your own photo
A free trial account gives you three searches a month, enough to see whether GeoPin can handle the kind of imagery you work with.