Sometimes one does have prior knowledge about the overall page layout of a document. For example, for historical monographs, there will often be no columns, but perhaps a few tables (which is, of course, very different, because the internal reading order of the paragraphs/cells in columns is vertical-first, but in tables it's horizontal-first).
It would be nice if there was a parameter (for the OCR-D wrapper) which can be used to constrain the results of the column detector. (A fixed number, or a range.)
Sometimes one does have prior knowledge about the overall page layout of a document. For example, for historical monographs, there will often be no columns, but perhaps a few tables (which is, of course, very different, because the internal reading order of the paragraphs/cells in columns is vertical-first, but in tables it's horizontal-first).
It would be nice if there was a parameter (for the OCR-D wrapper) which can be used to constrain the results of the column detector. (A fixed number, or a range.)