U-ViusalBERT --- Unsupervised Vision-and-Language Pre-training Without Parallel Images and Captions

Unsupervised Vision-and-Language Pre-training Without Parallel Images and Captions

2022-03-20 17:34:51

Paper: https://arxiv.org/pdf/2010.12831.pdf 

Code: https://github.com/uclanlp/visualbert 

1. Background and Motivation: 

本文拟解决 unpaired image-text data pre-training 问题。

=-=