Digikam

Useful settings

Collections -> Ignored Directories -> #recycle (for synology)

Metatdata -> Behaviour -> Use Lazy Synchornization: makes you confirm the changes before writing them below. Useful when working on lots of pictures (eg face recognition)

Metatdata -> Behaviour -> Face tags + geolocation information on

Metadata -> Sidecars -> write and read on.

Miscellaneous -> Behaviour -> Scan for new items at startup, fast scan, detect faces in newly added images. Fast scan will not detect items changed in external programs.

Miscellaneous -> Behaviour -> Initially select the first item in the album

On the bottom right, to the left of the zoom level, there is a little triangle with a status bar when digikam is operating. Press the triangle to see what it is doing.

Faces

In the preview you can select show face tags on and off in the top left of the preview picture

To remove a face, right click on a picture, remove tag and then select the person to remove

You can make a face a thumbnail by right clicking on a face in the People tab

You can see all the faces of a person in the people tab. You can see the whole picture in the Tags tab under People

To no longer scan for a person, you can right click the person in the People view, click Unmark tag as face, continue, but BE CAREFUL do NOT remove the tag corresponding to this face tag from the images! (so answer NO) You will then be able to find the tag still in the Tags list, but no longer in the people view.

To reset the training data (If you have edited a faces collection and removed people that don't belong), Tools -> Maintenance -> (Albums) -> face Engine management -> rebuild all training data

When scanning a large collection start with larger faces and high detection accuracy. Then scale it down and rescan, otherwise the amounts of hits will be huge.

If you move a person from a suggestion to Unkown, it will not suggest that picture for that person again. If you move it to Ignored, that face will never be suggested again.

To scan ignored faces

By design, digiKam won’t scan “Ignored” faces. To rescan them, you have to move them back to Ignored to Unknown.

Just an idea, but you might want to create a custom tag, somthine like “Temp Ignored”, and apply it to all the ignored faces before you move them back to Unknown. After you rescan, you can filter the Unknown view in the left people sidebar for your custom tag to easily move any previously ignored faces back to Ignored. Once the images are moved back to Ignored, I suggest you delete the custom tag, which will remove it from all the Ignored faces as well as any new matches.

The key point is to filter the “Unknown” people view with your custom tag after you run face recognition. If you use any other view it will include the newly tagged images, which means you’ll move both Unknown and new tagged images back to Ignored.

If you do this, don't run face recognition on the Temp Ignored view - it will find recognise people as being Temp Ignored and put them in there as suggestions.

Maintenance

Tools -> Maintenance -> Perform Database Cleaning

This should be done periodically to ensure everything is up to date and compresses the database. Digikam will stop responding during this cleaning as it doesn't want any conflicting inputs. This can take a long time!

Switching to Mariadb / Mysql

SQLIte (default) does not work too well over the network or on a NAS, so you may want to switch to Mariadb. This also allows every installation of digikam to access the same databases, so you can view the same faces etc on all your laptops and PCs.

I created a seperate database for each type (core, faces, similarity) for later scalability and put the thumbs on a local disk.

This Reddit thread gives really good instructions how to switch to Mariadb. More information on the official documentation here

Set the docker command to add

--max_allowed_packet=128M

You can check if it has worked by running the query

show variables;

Or by running

 mariadb-admin variables -p | grep packet

After you set up the new database you can migrate the data from the old sqllte over to the mysql using settings -> database migration.

I noticed not all the face data had been migrated, so under tools -> maintenance you can select sync metadata and database. NB make sure you change the direction to image metadata -> database. Also, a large collection can take a LONG time, so you can select which albums you want to sync. If you select an album with a lot of pictures in it, digikam will not respond for a long time. Just wait it out.