How to Organize Photos by Location When They Have No GPS

Why so many photos have no GPS

Location sorting depends on the GPS coordinates a camera writes into a photo’s EXIF metadata. The trouble is that a huge share of a real-world library never had those coordinates in the first place. Once you start looking, the gaps are everywhere:

• Screenshots — captures of maps, receipts, tickets and other pictures carry no camera GPS at all.
• AirDropped and messaged images — photos received from friends usually arrive with location already stripped.
• Scanned prints and slides — a scanner records the date you scanned, never where the shot was originally taken.
• Downloaded images — anything saved from the web or social media has typically been re-encoded with the metadata removed.
• Older cameras and phones — devices without a GPS chip, or with location logging switched off, simply never wrote coordinates.

The result is a library where date sorting works fine but location sorting seems impossible — because the one field every location tool reads is blank.

The usual dead end: date-only sorting

Faced with missing coordinates, most tools quietly give up on place and fall back to sorting by date. You end up with 2019 → 2019-08 → 2019-08-14 folders that tell you when but never where. That is fine for a timeline, but it is useless when you want to find “that trip to Chicago” or pull every photo from a single venue. A date-only tree buries a weekend in New York inside the same month as your kitchen renovation.

The AI-vision approach: read the photo, not just the metadata

When there is no GPS to read, DKPhotoSorter does what a person would do: it looks at the picture. Using the AI provider you choose (Google Gemini or OpenAI, on your own key), it examines the image for anything that pins down a place:

• Landmarks and monuments — a recognizable bridge, tower, arch or statue.
• City skylines — a distinctive downtown silhouette.
• Signs and storefronts — street signs, station names, shop fronts and building text.
• Venues — a named stadium, theme park, museum, hotel or restaurant.

Crucially, it doesn’t stop at a caption. Once the AI names a place, DKPhotoSorter runs geo-inference: it resolves that name to a real location and assigns a State/City (US) or Country/City (international) folder, verifying venues against OpenStreetMap so the folder names line up with everything sorted from true GPS.

Worked examples

• A photo of the Statue of Liberty with no coordinates → recognized as the landmark → filed under New York / New York City / Statue of Liberty.
• A shot inside a packed stadium → the venue is identified → the photo lands in that stadium’s city and a venue subfolder rather than a nameless date pile.
• A picture of a meal with the restaurant’s name on the window → the sign is read and the venue verified → filed under the correct city and restaurant.

Honest limits: it skips rather than guesses

Vision is powerful, but it isn’t magic, and pretending otherwise would scatter your library. DKPhotoSorter is deliberately conservative:

• Ambiguous scenes are skipped. A generic beach, a plain hotel room, a close-up of food with no context — if the photo could be almost anywhere, it is left out of the location tree instead of filed in a wrong city.
• Multi-city chains are skipped. When the visible clues point to several plausible places at once, the app declines to guess rather than pick one and be wrong.

Skipped photos aren’t lost — generic categories (Home, Family, Food and so on) stay in a flat structure, and non-photo files route to _Video, _Documents and similar folders. When you do know where an ambiguous batch belongs, you can help the AI: add Keywords to describe the context, or set a Tag City so those photos inherit the right place. You can also preview everything with Dry Run, and every real run is reversible with one-click Undo.

The same logic that catches a coordinate-free landmark also powers DKPhotoSorter’s everyday sorting. If you want the full picture, see how it organizes photos by location end to end, or step back to the guide to organizing thousands of photos for the complete workflow. Common questions are answered in the FAQ.

Private by design, on your own key

Vision analysis runs on the AI provider you pick, using your own API key. There are no developer servers, accounts, analytics or telemetry, and your original full-resolution photos never leave your computer — only a resized JPEG thumbnail plus limited metadata (filename, date, any GPS coordinates, the resolved place name, and, if you set them, your Home Address, Keywords and Tag City) is sent to Google or OpenAI. Gemini even offers a free tier for smaller batches, and every analyzed photo is cached, so re-runs cost nothing.