mmocr Integrate SVTR for Text Recognition

Integrate SVTR for Text Recognition

Open linhdb-2149 opened this issue 2 years ago • 6 comments

Describe the feature Paper Title: SVTR: Scene Text Recognition with a Single Visual Model Paper link: https://arxiv.org/pdf/2205.00159v2.pdf

Motivation This new method provide interesting approach to stroke-typed language such as Chinese/Japanese...

Related resources https://github.com/PaddlePaddle/PaddleOCR https://github.com/PaddlePaddle/PaddleOCR/tree/release/2.5/ppocr

Jun 30 '22 02:06 linhdb-2149

Hi @linhdb-2149 That's a good idea. To be done in the near feature.

Jun 30 '22 23:06 Mountchicken

@gaotongxiao @Mountchicken How can I contribute to this?

Oct 31 '22 12:10 bely66

@bely66 It's almost done. Thanks for your attention!

Nov 01 '22 01:11 gaotongxiao

@bely66 It's almost done. Thanks for your attention!

Thanks for letting me know

Nov 01 '22 09:11 bely66

@gaotongxiao Hi! Any news about SVTR? It's very interesting for me, because I can't reproduce paper's results by myself.

Dec 21 '22 17:12 trytolose

@trytolose Same here. We can't even reproduce the result with their original code. We will release SVTR by the end of this month anyway.

Dec 22 '22 00:12 gaotongxiao