the performance of the model

opconty / Transformer_STR

PyTorch implementation of my new method for Scene Text Recognition (STR) based on Transformer,Equipped with Transformer, this method outperforms the best model of the aforementioned deep-text-recognition-benchmark by 7.6% on CUTE80.

https://mp.weixin.qq.com/s/a_ahIwxiCaO7Bxmj81HUTw

222 stars 53 forks source link

the performance of the model #12

Open duxiangcheng opened 3 years ago

duxiangcheng commented 3 years ago

Hello, Great work from you, and thanks for the sharing. But I found that the best model only performed well on the cute dataset，what's your opinion on this? Could it be said that the encoder of the transformer is not good? Looking forward to your reply

opconty commented 3 years ago

Hi:

The shared weight pre-trained on Synthetic Dataset that we generated for curve text and other SynthText such as MJ/ST.

Maybe you should fine-tune this model on your own dataset if you yearning better performance.

Best.