Detecting Genre from Book Covers Using Deep Learning

I'm confident that the paper's title is the entire inspiration for this research: "Judging a Book By its Cover."

This is actually quite interesting. I used to work in a bookstore and as a productivity measure we would shelve books mostly without checking which section they belonged in. Turns out it is pretty easy to tell based only on the cover.

I'm not trying to be a luddite, and it's all fun, but to me this seems like a particularily bad idea.

Sure, this obviously works most of the time, especially on romance novels. But the end result if actually used anywhere is that books would be either made to look as homogenic as possible, or they would be mischaracterized when being artistic in a nod to some content in the book.

Judging a book by its cover doesn't work with humans, and is a bad idea, ergo it's still a bad idea to have a machine do it

Or maybe covers are pushing clichés too much

Same with movie posters (Big Face over scenery, couple standing opposite to each other, Papyrus font, etc)