Measurement unit in PDF's XML output
The size in the XML (data.xml) generated by pdftohtml for this PDF (input.pdf) is 595x841. The pdfinfo utility also prints the page size as 595x841 pts.
Found online that 1pt is 1.33px. However, that does not seem to apply here. The image
data-1_1.png's top is 27pts. But when checking the distance in pixels in the PDF, it is not ~35px (27 * 1.33) but ~55px.