David You are right, that there are scenarious, when it makes sense to offer suggestions in large batches. When you start marking up faces for the person - you have to start small. When you almost finished - you also have to very carefull. However, if you are in the middle of the process, there are still thousands of images with that person, you might go with bigger chunk. The challenge is though, program cannot know for sure how many of remaining faces are actually belonging to this person. Otherwise it would not need person to approve anything 🙂
However, I think it is possible to make some sort of smart threshold that would estimate accuracy of latest accepted suggetsions and do some assumptions where we are in this process - in the middle or not, and how many correct faces can there be. This is interesting scientific task.
However, simple solutions like “let users decide the batch size” will not work. Since for one person it makes sense to provide more suggetsion, but for another person in the same library that will only result in huge amount of garbage suggestions. So there needs to be more delicate approach.
