Glossary · Similar-photo grouping
What is similar-photo grouping in a cleaner app?
Short answer
Similar-photo grouping is the step where a cleaner clusters photos that show the same scene, taken close together in time, so you can keep one and remove the rest. It combines a visual similarity check with capture time. Unlike duplicate detection, it finds different files that look alike, such as five frames of one pose.
Last updated
How does grouping differ from duplicate detection?
| Exact duplicate check | Similar-photo grouping | |
|---|---|---|
| Finds | Byte-identical copies of one file | Different frames of the same moment |
| Method | File hash comparison | Visual fingerprint plus capture time |
| Safe to auto-delete? | Yes, if one copy stays | No, someone must choose the keeper |
| Found by free tools? | Usually | Rarely |
A near-duplicate photo is the unit that grouping works with: each group holds a set of them plus the frame that will be kept.
What steps does a grouping engine follow?
- Read each photo's capture time from its EXIF metadata.
- Sort photos into time windows, often a few seconds to a couple of minutes apart.
- Compute a compact visual fingerprint for every photo, such as a perceptual hash or an AI embedding.
- Within each window, join photos whose fingerprints are close enough.
- Rank the photos inside each group and mark one as the keeper.
Using time first keeps the work fast and avoids absurd matches. Without it, every photo of a white wall or a blue sky could land in one huge group. It also respects how people think: a burst is one moment, and the same view next summer is a new memory.
Where does grouping go wrong?
- Too loose: two different poses at the same table get merged, and one is offered for deletion even though both matter.
- Too strict: a burst where the camera panned slightly splits into several small groups, leaving copies behind.
- Edited copies: a cropped or filtered version may or may not join its original, depending on the fingerprint.
- Missing timestamps: images received through chat apps often lose their original capture time, so they group by visual match alone.
Because of these edge cases, a group should always be shown side by side before anything leaves the phone. Being able to drag a photo out of a group, or mark a second keeper, matters more than how clever the matching is.
Which tools group similar photos?
The iPhone Photos app's Duplicates album merges exact and near-exact copies but not ordinary bursts. On Android, Files by Google focuses on exact duplicate files. Dedicated cleaners add true similar-shot grouping. GalleryJury, an Android cleaner that is not yet on Google Play, groups each burst, keeps its strongest frame and leaves favourites and protected albums untouched. For a manual route, see how to delete similar photos on Android.
Frequently asked questions
How close in time do photos need to be to group?
It varies by tool. Seconds catch bursts; a few minutes also catch repeated attempts at the same group photo, such as a family lining up three times. Very long windows risk merging separate moments.
Will grouping catch a photo I took again a week later?
Usually not, and that is deliberate. A cake photographed at two birthdays is two memories, so time limits keep them apart. If you want old repeats found too, a separate duplicate check on file content handles exact copies across any date.
Sources
- Files by Google Help: Delete duplicate files (checked 31 October 2026)
- Google Photos Help: Manage your storage (checked 31 October 2026)
Related
More in Photo cleanup terms
GalleryJury is coming soon to Android
It keeps the best shot from every burst and clears the clutter around it. 50 free photos to try, then $5 for 5,000 photos or $15 for 20,000. Pay once, nothing renews. No subscription.