In this article, you will learn how people are identified automatically.
In order for the AI module person identification to work reliably, it must be trained with two different images showing the face of the specific person. Please use high-resolution photos (without visible image noise) that show the face frontally and without headgear.
After uploading, open one of the images in large view > a frame will now appear around any faces found and the + Assign person button will appear underneath > if you click on this, a list of all persons created will appear (a new entry can also be made at this point if necessary) > finally, click on the name and repeat the process for another image.
If you now load new images into the Media Hub or re-index the AI data of existing folders (see here), the person will be reliably recognised and automatically tagged. In rare cases of uncertainty, the AI suggests possible persons for manual decision.
Resetting the AI training
If you want to reset the AI training for a specific person (for data protection reasons or to restart the training), simply click the round arrow symbol ↺ in the person management and confirm the deletion dialog.
💡 Tip: If face recognition is not possible for one or more images, e.g. because faces are partially obscured, you can of course always search for created persons in the normal tag mask as a workaround and add the desired person tag to the corresponding images manually.