I wrote a package "impocr" (https://github.com/laurent-pincemaille/Impots-ocr) that is intended to read and extract (ocr) some information based on pdf notice of a