We are building a stock image search engine. And we want it to be the best search tool for stock image users ever. At the moment we can search across hundreds of millions of stock images and unfortunately not all of them are great. Far from it. And this is where our aesthetic detector comes in handy. There are two very important tasks it can accomplish:
Neural network would estimate a visual quality of every image and apply aesthetic score to every file. Later on this data would take part in the overall mix of ranking factors and help improve search results by bringing aesthetically better images to the first pages.
There are lots and lots of unforgivingly bad stock photos. Neural network would detect photos with the lowest aesthetic score and literally weed them out from the search results.