I once built a prototype app to soft delete near-duplicate photos to save space. i.e. it would look at images that are very similar and identify the one that was sharpest, with the best lighting etc, and soft-delete the rest. The app worked but was very slow due to Python's image processing libraries. But if anyone here thinks this is a good idea happy to share how I did it as I'm sure a competent engineer can make this a reality.