Best practices for massive volume image QC

I am designing a system that will be used to perform a visual QC step on literally hundreds of thousands of images per day. We have already taken care of the obvious to spot problems with AI, reducing human intervention. However, I anticipate this making the task even more boring and difficult for humans to spot minor issues with images. I’m trying to increase productivity while not making this task unbearable.

The old system this was replacing had people scrolling a page at a time through pages of about 30 images at a time, spending lest than a second per page. This was fine before our volume started increasing but this doesn’t scale. Without getting too specific, we are trying to weed out poor quality images, not censor or choose the best among them.

Anyway, I’m trying to find other applications that solved a similar problem and learn how they added value to the process with UX. I’m thinking something like used by the poor souls who have to manually screen images posted on social media or image mapping applications like google street view. Either this is difficult to describe to search engines or proprietary or secretive so I’ve already tried before coming here.

If anyone has worked through this type of problem before, I’d appreciate you sharing some insights or pointing me toward the right people or sites.