search

mjq11302010044 / TATT

A Text Attention Network for Spatial Deformation Robust Scene Text Image Super-resolution (CVPR2022)
MIT License
167 stars 17 forks source link

Pre-trained model #1

Closed j-lootens closed 2 years ago

j-lootens commented 2 years ago

Would it be possible to make a pre-trained model available?

mjq11302010044 commented 2 years ago

@TheNoodleNinja Sorry, we don't actually need a pre-trained model, you can directly train the model on the TextZoom dataset.

j-lootens commented 2 years ago

Np but it might be useful for people who only have limited VRAM or no dedicated GPU at all.

mjq11302010044 commented 2 years ago

@TheNoodleNinja If you mean the model trained TextZoom, that will be added to the release. Thanks for your advice though. :)