wangyuxin87

[This issue](https://github.com/wangyuxin87/VisionLAN/issues/10) may help you.

python demo.py

Thanks for your attention, you can easily reimplement PERT following our published paper. The official implementation will be available after the paper being accepted.

It seems that you do not correctly configure your environment. Please examine if you ignore some steps of our proposed step-by-step installation. In addition, our enviroment is shown as following:...

We use Image editor (e.g. photoshop) to manually occlude the character.

Thanks for pointing out this issue, It works now.

G的形式为Eq.3。 1.我们简单地进行了2个字符遮挡的实验，效果有提升。 2.在我们的实验中，MLM可视化结果更关注的是位置信息，例如重复的字符中，MLM能准确根据位置定位而不是内容定位。 3.单个字符不需要语言信息。

Thanks for pointing out this problem. It works well now.

Thanks for you attention. As we just posted it a few days ago, it will be retrieved in a few days

The following one is correct: @article{wang2021two, title={From Two to One: A New Scene Text Recognizer with Visual Language Modeling Network}, author={Wang, Yuxin and Xie, Hongtao and Fang, Shancheng and Wang,...