Question/development request: create PersonMarker from PersonInImage tag

[Uwe]

Hello,
I have not yet delved deeply enough into IMatch. Therefore, here is the following status and the development request that may result from it. So if there is already a possibility, then I would be very happy about an existing solution.
The photo stock comprises approx. 250000 photos, which I have managed in various DAMs over the last 20 years. What they all had in common was that I could enter which people were depicted in the photo (stored in the XMP-iptcext:PersonInImage tag but without the XMP-mwg-rs region tags). In other words, I know who is in the photo, but without the ''marker''. Now I don't want to adjust all photos again to generate the 'region tags'. Is there a way in IMatch to read the PersonInImage tag and generate 'dummy markers' with the names from the PersonInImage tag in the viewer for each person stored there, so that I only have to drag the 'green square dummy marker' onto the person. If the person has already been edited, then I can delete the dummies straight away. Of course, the current PersonInImage tag also contains names of people whose faces cannot be determined via face recognition, e.g. because they are only partially recognisable, have turned away, etc., but it was important for me to enter the person. I would place the 'dummy tag' somewhere on the person, for example.
Uwe

Translated with DeepL.com (free version)

[Mario]

IMatch has a special feature for exactly this purpose.

Maybe Importing PersonInImage Values will help. This off by default, you need to enable it before importing the images as explained in the help topic. IMatch uses the PersonInImage tags and creates links. See Linking Persons to Files

If you already have imported the images into your database, you will have to re-import the metadata again with a forced import: Select files in a File Window, press Shift+Ctrl+F5 > Reload Metadata. If the files have pending metadata, write back before.

[Uwe]

Hi Mario, thanks for your fast response. Here are my settings of Metadata2. What do I have to adjust because the Shift-CTRL-F5 does not generate the "blue boxes" with the correspondig name existing in the PersonInImage tag.
Regards, Uwe

[Mario]

Settings look OK. Provide a sample image + XMP sidecar for analysis. Upload to cloud and post a link or send to support email address with a link back to this thread (I get a lot of emails every day).

Just made a test with a JPG with embedded PersonInImage and a CR2 with XMP sidecar. Using your settings imported PersonInImage and created links to the persons in the image (blue link annotations).

[Mario]

I have downloaded your sample images. All are loaded, PersonInImage is converted into person links, works as intended.
More details in the E-Mail I have sent to you today.