Datenmodellierung

[blrtvs]

[maxnth]

From my understanding the proposed data model only works with the PAGE XML files where text regions have a maximum of 4 coordinates which isn't always the case (e.g. when using polygons for page segmentation in tools like Aletheia or LAREX).

I attached an example file with more than 4 coordinates which was produced by OCR4ALL.
0006.zip

[maxnth]

We should also keep in mind that a page can have more than one "TextRegion" element and that the Reading Order can play an important part in analyzing the extracted text.