How Many Images Do You Need to Retrain a Model?

Tori: What's the amount of data that is significant enough to actually change the results of a model? I assume Tori's going to need a few hundred devices. Well, so the number of images that you'd require to retrain the model is very much dependent on the number of labels that you get on the other end. It seems like it's a much more discrete data set, which is probably easier for it to learn to distinguish, right? And if there was a continuous, very organic growth of these devices where every phone was really unique. Yeah. If I had just one instance of each device, it would probably still do a reasonable job I would think.