Could i train/ fineturn donut for text direction task?

clovaai / donut

Official Implementation of OCR-free Document Understanding Transformer (Donut) and Synthetic Document Generator (SynthDoG), ECCV 2022

https://arxiv.org/abs/2111.15664

MIT License

5.53k stars 444 forks source link

Open chopin1998 opened 12 months ago

chopin1998 commented 12 months ago

i have lots of photo with text, i need to class them by 0/ 90/ 180/ 270 by text angle?

can donut do this task?