How to Identify a Landmark From a Photo

Most “what landmark is this” questions take seconds to answer. Point a recognition tool at a cathedral, a bridge or a famous tower and it returns a name and a few facts. It works by comparing your picture with images of places it already knows — Google describes Lens as ranking other images “based on their similarity and relevance” to yours. A well-photographed landmark matches. An ordinary building gets look-alikes, or sometimes the name of a more famous building in the same style.

That is a different job from reverse image search, which hunts for other copies of the same picture — the engines compared covers that side. This guide is about getting a name: which tools give one, how to use each, and what to do when the building is too ordinary to be on file.

The order matters less than the mix. Lens, Visual Look Up, Bing and Yandex come from four different companies with different data behind them, so a building one misses can still turn up in another. Two agreeing answers from different companies beat one confident answer. For what Lens can and cannot tell you about where a photo was taken, see whether Google Lens can find a location.

1. Step 1On a computerGo to google.com, click Search by image, then Upload a file and choose the photo. In Chrome you can also right-click an image on a web page and choose Search with Google Lens. Google says Lens works in most browsers, including Chrome, Firefox, Safari and Edge.
2. Step 2On AndroidOpen the photo in Google Photos, tap More at the top, then Google Lens.
3. Step 3On iPhoneIn the Google app, Chrome or Safari, tap Google Lens in the search bar, tap the photo picker and select the photo. From Google Photos, tap More, then Google Lens.
4. Step 4Select just the buildingDrag the corners of the selection box so it frames the landmark. Google’s own tip is that a smaller area gives more specific results. Keep the top of the building inside the box, because the roofline, spire or dome is usually its most recognisable part.

1. Open the photo or image full screen. Visual Look Up works in Photos, Safari, Quick Look and other apps. In Notes, Mail or Messages, touch and hold the image to select it first.
2. Check the Info button. A symbol with stars means Visual Look Up has something for this photo. With no star, it is not available for that photo.
3. Tap the starred Info button, then tap Look Up at the top of the photo information.
4. Read the result, which Apple draws from Siri Knowledge and the web, and tap it to learn more.

Apple’s requirements are iOS 15 or later on iPhone, and iPadOS 15.1 or later on supported iPads. Language matters as well. At the time of writing, Apple’s feature availability page lists Visual Look Up in 14 languages: English, French, German, Italian, Spanish, Turkish, Japanese, Korean, Thai, Dutch (Netherlands), Portuguese (Brazil), Arabic (Saudi Arabia), and Traditional Chinese for Hong Kong and for Taiwan. If the stars never appear on any photo, check that list before you blame the picture.

It is separate from visual intelligence on Apple Intelligence iPhones, which Apple presents as a camera tool — for example, to look up a restaurant or business in front of you. For a photo already in your library, Visual Look Up is the direct route.

Bing Visual Search. Microsoft’s page says you can upload an image to “identify objects, plants, animals, landmarks, and more”. On Windows it is unusually close to hand. In the Photos app, open a picture, click the three dots and choose Visual search with Bing. In Edge, right-click and pick Visual Search to frame part of a page. Microsoft notes that availability varies by device type, market and browser version.

Yandex Images. At yandex.com/images, click the image search button to the right of the search bar. Then drag in the photo, choose a file, paste it or enter its web address. Yandex says the search can tell you what is in a photo and find similar images. It also warns that results depend on whether the images exist on the internet and have been indexed. In practice, it can only return a building that someone has already photographed and published.

Which engine tends to win on which kind of picture is covered in the Lens vs Yandex comparison.

Send an image to the Vision API with the feature type LANDMARK_DETECTION. Google’s documentation says it “detects popular natural and human-made structures within an image”. Each result carries the landmark’s name, a score from 0 to 1, its position in the frame and its coordinates. Leave out maxResults and you get the default of 10 results.

Billing is per unit, and each feature applied to one image counts as one unit. Landmark Detection is free for the first 1,000 units a month. After that it costs $1.50 per 1,000 units up to 5 million, and $0.60 per 1,000 beyond that.

Read the coordinates with care. Google’s API reference says a result can hold more than one location: one for the scene in the image, and possibly another for the place the photo was taken. A tower can be photographed from kilometres away, so the landmark’s position is not proof of where the photographer stood. Closing that gap is a job for Street View. For APIs that estimate a region from any outdoor scene, not just a famous one, see the photo geolocation API guide.

A block of flats, a village church or a street corner rarely has a name that a recognition tool would know. The question changes from “what is this building?” to “where was this photo taken?”, and that one is answered in a different order:

1. Step 1Check for GPS data firstIf the original file still has its metadata, the coordinates are already inside it. Reading them takes a minute, and a measured position beats any estimate.
2. Step 2Narrow the region from the cluesThe script on signs, the side of the road traffic uses, road markings, roof shapes and plants all narrow a photo down. The clue checklist sets out which to read first, and shadows can settle the hemisphere.
3. Step 3Get a city-level estimateGeoSpy estimates the city and country from what is visible — architecture, vegetation, signage, road markings and light — and reads EXIF GPS when the file still has it. The result is a city-level estimate, not an address, and uploaded photos are not stored.
4. Step 4Search the area yourselfWith a city in hand, look for the building in Google Maps 3D and Street View, as described in the next section.

On the web, photo analysis comes with a GeoSpy Premium subscription, weekly or yearly, which renews until you cancel. The GeoSpy iPhone app has a free trial. How a model turns clues into a region is explained in how AI geolocation works.

1. Open Google Maps in Satellite mode. Hover over Layers, click More and tick Globe view, then click the 3D icon at the lower right, below the compass.
2. Tilt and rotate the view. Height, roof shape, domes and towers are much easier to compare at an angle than from straight above.
3. When a building looks right, click Pegman at the bottom right and drop it on a blue line, blue dot or orange dot nearby.
4. Move until the view matches your photo, then click See more dates to compare older captures. Google notes that historic imagery is not available for every place that has Street View.

Street View is dense in some countries and thin or missing in others — the coverage guide shows where. Where it is missing, satellite imagery still shows roof shapes and how the streets are laid out.

• Match the angle. Find where the photographer stood. If a spire lines up with a rooftop corner in your photo, it must line up the same way from that spot; a few metres to either side and the overlap changes.
• Count floors and window bays. Buildings in the same style rarely share the exact number of storeys, windows per floor, balconies and doorways. Count them in both images.
• Read the signs. Shop names, street signs and plaques can be searched directly or typed into Maps. Text is the most specific clue a photo can hold.
• Compare old and new imagery. Facades get cleaned, repainted and rebuilt, and signs change. A mismatch can be a difference in date rather than a wrong building, so check older Street View captures and archive photos.
• Rule out look-alikes. Famous buildings have replicas and imitators — Google’s own explanation of Lens uses the Eiffel Tower as its example. Check the skyline, terrain and neighbouring buildings, not just the landmark.
• Check for a mirror image. Reposted photos are sometimes flipped. If lettering reads backwards, mirror the picture before comparing.

The same discipline applies whenever you verify a photo: a match you cannot explain detail by detail is still a guess. For a fuller version of these checks, see the OSINT geolocation workflow.