I have two suggestions to improve the workflow on face detection.

1. Tab-layout on Detecting/recognizing faces:
Problem: There are two workflows, started with the same button.
The options are scattered about the tabs. It is unclear which option/setting works for which workflow. Example:
It is not clear whether the 'accuracy' slider is about detecting faces, recognizing faces, or both.

Suggested improvement: it would be better to have 2 tabs: One with the settings for 'detecting faces' with a button 'detect faces now';
another with settings for 'recognize faces' and a button 'recognize now'.


2. How to tell Digikam which face is which person:
This might be a big change, but hear me out:

Right now the software needs a name to a face ('this is Julia'), finds suggestions ('Digikam thinks this is Julia') and needs confirmation.

My Synology NAS also does face recognition and it works very smoothly.
It groups a whole lot of similar faces without knowing the name.
It says 'I think these 38 faces are all the same person - please provide a name'.
That group of people is displayed as one thumbnail and the number of faces in the group.
It's a different workflow and it works very fast.