Consider extracting the fields first, then submitting them to tesseract
separately. There is no guarantee tesseract will return the fields in
order, also lines confuse it.
Try enlarging the image 100%, that's about the ideal letter size for
Tesseract.
On Sat, Dec 21, 2024 at 6:56 AM Riccardo
wro
other example:
https://www.kaggle.com/code/sreesankar711/table-transformer-demo
Zdenko
so 21. 12. 2024 o 19:37 Zdenko Podobny napísal(a):
> Hi,
>
> have a look at this example:
> article:
> https://iamrajatroy.medium.com/document-intelligence-series-part-2-transformer-for-table-detection-extr
Hi,
have a look at this example:
article:
https://iamrajatroy.medium.com/document-intelligence-series-part-2-transformer-for-table-detection-extraction-80a52486fa3
notebook:
https://nbviewer.org/github/iamrajatroy/Data-Science-Lab/blob/main/notebook/DETR_Document_Intelligence.ipynb
Zdenko
so 21
3 matches
Mail list logo