J'ai utilisé l' AI pour générer les infos sur les bugs de reconnaissance des personnes

Cela ne représente pas totalement mon opinion, le logiciel est performant sur un certain nombre d'éléments dont les albums et la gestion des étiquettes

Voilà le texte :

Bug Report Draft (English)
Title: digiKam 8.9 – Face validation in batch blocks People management

Version: digiKam 8.9 Platform: [specify OS, e.g. Windows 11 / Linux distribution] Face recognition model: Slowest model (most precise)

Steps to reproduce
Run Maintenance → Rebuild faces database (after noticing faces were mixed).

Go to the People view.

Select multiple detected faces.

Validate one of the selected faces (all selected faces are validated at once).

Try to manage people afterwards (rename, merge, edit, etc.).

Actual behavior
All selected faces are validated in batch.

After this operation, People management becomes almost blocked:

The interface is unresponsive or extremely slow.

Actions on individuals (merge, rename, edit) cannot be performed.

The background process seems to continue, but the UI is unusable.

Expected behavior
Batch validation should work without blocking the People view.

After validation, it should remain possible to manage individuals normally.

Impact
Recognition and validation of faces, which should accelerate organization, instead causes database inconsistencies and workflow disruption.

This severely limits the practical interest of digiKam’s face recognition feature.

Notes
The issue seems related to database integrity after batch validation.

Using the slowest recognition model may exacerbate the problem.

Workarounds tested:

Running Maintenance → Clean database and Rebuild fingerprints

Validating faces in smaller groups or individually

These reduce but do not eliminate the issue.