Muharaf Dataset with Standard PageXML Use

This deposit contains a modified version of the public part of the Muharaf dataset. The original distribution's PageXML files used a proprietary center-line element to encode the baseline of for each text line's bounding polygon. In contrast to the baseline element provided by the standard, the polyline encoded by this element changes direction depending on the reading direction (LTR/RTL) of the text on that line. This has been corrected to conform more closely to standard PageXML practices with baseline elements whose polyline is defined in the left-hand side is up format.