Dense Contrastive Visual-Linguistic Pretraining

Lei Shi, Kai Shuang, Shijie Geng +5

Inspired by the success of BERT, several multimodal representation learning approaches have been proposed that jointly represent image and text. These approaches achieve superior p…