OCR PDF Structure Issue

[bikramnayak]

I’m working with a scanned PDF that contains a table with two columns, where each column has two lines of text. When I convert the scanned PDF using OCRmyPDF, I’m encountering an issue with the resulting content. Tesseract processes the text line by line, but this causes OCRmyPDF to generate separate spans for each piece of content. Specifically, it creates a span for row 1, cell 1, then another span for row 1, cell 2, followed by separate spans for row 2, cell 1, and row 2, cell 2. This results in accessibility problems for screen readers, as the content is not structured properly. Is there any way to resolve this issue and ensure the table is interpreted correctly by screen readers?

[jbarlow83]

We depend on Tesseract or the OCR engine to generate usable information. I think one would get better results from improving the OCR engine than trying to repair the layout at OCRmyPDF's level.