Mejoras en el OCR Tesseract

Abstract

The Library of the University of Seville owns a rich collection of antique works that are being scanned. In order to make the access to information easier, they need the full text of these books. An automated transcription solution is sought since the task is too time-consuming for humans. To achieve this, different existing solutions are analyzed, such as veteran open source programs and cutting-edge neural networks technologies. After the research, a concrete solution is provided to the Library, along with guidelines for its execution.