mjq11302010044 / TATT

A Text Attention Network for Spatial Deformation Robust Scene Text Image Super-resolution (CVPR2022)

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

请作者开放预训练模型

zhuxyme opened this issue · comments

@zhuxyme TATT has no pre-trained model, you can directly train from scratch. For the TP Generator, you can download weights following ReadMe.md and perform the initialization.

I think what he calls pre-training parameters refers to the completely trained model parameters that can get the results in the paper.

TATT has no pre-trained model, you can directly train from scratch. For the TP Generator, you can download weights following ReadMe.md and perform the initialization.

YES

是的

@zhuxyme Please send your request to our email: mjq11302010044@gmail.com and provide your institute name and research purpose, we will provide you the trained parameter.