poppler-science

Abstract

The “Poppler-science” software is a fork of the existing open-source Poppler project (https://poppler.freedesktop.org/) for converting PDF files to text. Modifications to the Poppler source code include (a) per-glyph optical character recognition (for correcting the non-standard font glyph remapping that is common in the scientific literature), (b) inference of text markup for commonly used scientific formatting (like superscripts and subscripts), (c) table and figure recognition, and (d) improved ordering of text output for complex scientific manuscripts (e.g., multi-column text, figure and table captions, etc.).