Fast, Diverse and Accurate Image Captioning Guided By Part-of-Speech
Fast,DiverseandAccurateImageCaptioningGuidedByPart-of-Speech原文地址时间:2019年概括本文致力于建立一个fast,accurate,diverse的imagecaption模型,首先提出了传统beamsearch的慢、diversity不足,以及基于GAN、VAE方法的准确率不足的缺陷,然后提出了根据词性序列生成caption的方法,这