Immich.app as media viewer with a good face recognition for my Nextcloud server

Photos and Recognizes are not really suitable, since Recognize detects way too little faces. The older face recognition app has a great detection functionality, but needs manual reorganisation after new photos are uploaded. The classification part is not that great. So I try my luck with the Immich.app, integrated into the Nextcloud structure as media viewer with great face recognition and other excellent features.

Here a short howto, which is not perfect yet, but works quite good:

Interesting, does it work with Object Storage as well or just local disks?

Iā€™m having troubles understanding the gist of your tutorial: Is it how to use external storage?

Did not try it with object storage yet, might be worth a test.

The main points are to run the docker as user www-data and mount the nextcloud data/user folders as immich storages to use them as an external library there.

Just tested the Immich.App as a substitution for Nextcloud Photos/Recognize/Machine learning (Object detection) and Facerecognition. Immich even supports OCR which can also read text within your pictures. It is faster, more intuitive, everything just works like a charm. It even processes videos (also for face detection) and supports really all formats, since I photograph a lot with Sony, Fuji, Canon and Nikon RAW files, TIFS, iPhone live photos, h265 hevc, heif, whatever. Immich just shows everything without any errors. Recognize does not even start its background jobs anymore since I deleted those and wanted a reclassify. Also not within the occ CLI. I updated my link above with tests and numbers.

1 Like