how to update the pretrained model with new vocabulary to improve the recognition part

princewang1994 / TextSnake.pytorch

A PyTorch implementation of ECCV2018 Paper: TextSnake: A Flexible Representation for Detecting Text of Arbitrary Shapes

https://arxiv.org/abs/1807.01544

MIT License

437 stars 92 forks source link

how to update the pretrained model with new vocabulary to improve the recognition part #60

Open naarkhoo opened 3 years ago

naarkhoo commented 3 years ago

In my case TextSnake works well on detecting the text region however the recognition part is not so great - this is mainly because the domain text is medical. I have synthesized text images with custom library but I wonder if it is possible to update the pretrain model instead of training a totally new model from scratch