Bug 514139 - [Feature request] use existing tags in DB as retrieval-augmented generation or tuning material when suggesting new tags for images
Summary: [Feature request] use existing tags in DB as retrieval-augmented generation o...
Status: REPORTED
Alias: None
Product: digikam
Classification: Applications
Component: Tags-AutoAssignement (other bugs)
Version First Reported In: 8.8.0
Platform: Other Other
: NOR wishlist
Target Milestone: ---
Assignee: Digikam Developers
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2026-01-04 11:33 UTC by sean dmr
Modified: 2026-01-04 11:58 UTC (History)
1 user (show)

See Also:
Latest Commit:
Version Fixed/Implemented In:
Sentry Crash Report:

Attachments
Add an attachment

Note You need to log in before you can comment on or make changes to this bug.
Description sean dmr 2026-01-04 11:33:39 UTC 
SUMMARY

Right now, auto-tagging is limited to objects (person, dog, car).

I also use more "thematic" tags, like "winter | winter light" or "alone" or "couples", etc. I imagine this is different for every user.

It would be fantastic if digikam could auto-suggest tags for new images based on my historical tagging. Things like:

- image content (e.g. "alone", "couple")
- exif data ("winter", "nyc | soho")
- combos (exif: "winter"; content: "winter light")
- etc

SOFTWARE/OS VERSIONS
digiKam: 8.8.0
KDE Frameworks: 6.17.0
Qt: Using 6.9.1 and built against 6.9.1
macOS Sequoia (15.6)
Build ABI: arm64-little_endian-lp64
Kernel: darwin 24.6.0
Comment 1 caulier.gilles 2026-01-04 11:54:40 UTC 
What do you mean by "RAG" ?
Comment 2 sean dmr 2026-01-04 11:57:53 UTC 
(In reply to caulier.gilles from comment #1)
> What do you mean by "RAG" ?

Retrieval augmented generation, if it made more sense to do the primed inference at runtime rather than ahead of time with tuning.