How to manage face recognitions of people from the crowd?

Hi, I’m using recognize and memories apps from a couple of months now, using face recognitions and clustering on a repository of around 50K pictures. It works pretty well! Thanks!

What would be the good way to manage “people from the crowd” ?

For now, when I manualy review the pictures of a selected known cluster/people, I move pictures with unknown people to a dedicated cluster/people named “crowd”.

Does this practice not risk biasing the clustering function which would end up associating every picture to this “crowd” cluster/people?

Thanks again for this awesome work!

1 Like

Yes, you are right. Having a cluster encompassing all sorts of facial features is bound to cause problems.

1 Like

Thanks!
So, is there any way to avoid some faces to be clustered?

I don’t think this is possible, currently. Sorry

Ok, thanks! Anyway great job!