[discuss] approaches for reading order detection

Beginners
1

what are the different models and dataset using ML techniques to find the reading order of layout sections in pdfs? any open license model will be a plus

2

To find the reading order of layout sections in PDFs using ML techniques, you can explore models and datasets like:

  1. LayoutLM: Uses transformer models to understand the layout and reading order in documents.
  2. DocFormer: Focuses on document understanding and layout analysis.
  3. OCR and Layout Analysis Datasets: Open datasets like PubLayNet and DocVQA can be used for training models. locksmith services

These models and datasets can help identify and process reading orders in PDFs.

1 Like
3

Surya is a document OCR toolkit that does:

  • OCR in 90+ languages that benchmarks favorably vs cloud services
  • Line-level text detection in any language
  • Layout analysis (table, image, header, etc detection)
  • Reading order detection

There seems to be a HFH Space running it: Surya OCR - a Hugging Face Space by artificialguybr

1 Like
4

Thanks for the response. This seems to have some restricted license. Anything with open use policy? Found one but still integration is not straight forward; HURIDOCS/pdf-reading-order · technical details about the features used? 2 models?