Anyone figured out how to handle long text sequences

[bely66]

All the recognition models can't handle long sequences

For some reason, it gets a bit better however it never matches the accuracy of single words

and most of the papers say they can handle long sentences with high accuracy

I tried all the models (CRNN, SAR, ABINET, SVTR)

Has anyone figured it out, yet?
is there any trick with the data preprocessing? Does anyone have some ideas?

Thanks everyone

[bhavyajoshi-mahindra]

I am also looking for the same answer.
My paddleocrV4 finetuned model is able to correctly recognize single words but is not able to recognize long sequence of words in a image.

[milosacimovic]

Any update on this?

[UserWangZz]

At present, en_ppocrv4 only open-sources the mobile model, which is also optimized for long texts. However, it still shows problems such as insensitivity to space prediction, and easy recognition of brackets and punctuation. In the future, we will open source the en_ppocrv4_server model, which effectively solves the above problems of the mobile model and achieves 91% accuracy in the internal test set. Compared with the 77% accuracy of the mobile model, it is an increase of 14%.