Still vastly underused in image to image to this day and I don't understand why: Unets and self-attention. I keep seeing papers come out that have obvious problems that could be solved using these two simple things!
-
Care to elaborate?
2/ And then there's Unet's- it's just a plain brilliant and simple way to get a big head start on image to image, because most of the legwork is in recognizing objects. Furthermore, acknowledging that you'll probably not match Google or Facebook's vision models on this.
