Split extend Dataset to known<>unknown or more detailed

[ChrizZz90]

This is just an idea to enhance the extended dataset feature.

Today you can only extend the dataset or you can skip this feature. My idea is to be more detailed. At least to decide between known or unknown faces or to create a list based on the names to define which dataset should be extended.

For example, John and Jane are living in the house where the mirror is placed. Today their dataset will grow heavily. After a while, they don't need to extend their dataset. But if friends are visiting them, they have maybe just a dataset with less than 10 pictures. For them, it is useful to extend the dataset.

What do you think about this idea?

[nischi]

But why you wanna have multiple datasets? You can put your friends in the main dataset. For each face you have a own name. And per name you can decide what to show. Or do i miss something here?

[ChrizZz90]

The idea is more about the decision if I want to extend the dataset or not. Right now I extend the dataset and thousands of pictures are made with my face. This isn't necessary. But for friends with a pool of just a few pictures, it makes sense to extend the dataset with new pictures and to make the dataset more accurate. It is still 1 dataset, but the pool of images how the dataset is created differs. Based on this feature you could improve your data pool just if it is a real improvement.

[nischi]

I think i know what you mean now. So basicaly you need a setting where you can tell the system which person you wanna extend. Means only this photos will saved for dataset extensions, and the others will be "ignored". Correct?

[ChrizZz90]

yes, exactly

[eurtnl]

Maybe to add a counter per known person: when there are 25 pictures for a known person stop extending it.