(Paper Reading)Bottom-Up and Top-Down Attention for Image Captioning and Visual Question Answering
IntroductionWithinourapproach,thebottom-upmechanism(basedonFasterR-CNN)proposesimageregions,eachwithanassociatedfeaturevector,whilethetop-downmechanismdeterminesfeatureweightings.Inthispaperweproposea