Extract text from an image called filename.jpeg using tesseract within a docker container. The image contains german text, therefore `-l deu`. The text regions or zones are defined by `filename.uzn`. The image filename and the .uzn filename have to be identical for this to work!

create_uzn.rb

Dir['./books/*/*/*.jpeg'].each do |file_path|
  next if File.directory? file_path 
  file_name = file_path.sub('.jpeg', '.uzn')
  puts "OCR #{file_name}"

  uzn_content = """2385 1059 273 201 Energie
2634 1080 372 210 Kohlenhydrate
2961 1062 294 234 Fett
3204 1080 321 186 Eiweiss
2030 1500 800 960 Zutaten
2823 1497 975 1023 Zubereitung"""

  File.write(file_name, uzn_content)
end

filename.uzn

2385 1059 273 201 Textblock 1
2634 1080 372 210 Textblock 2
2961 1062 294 234 Textblock 3
3204 1080 321 186 Textblock 4
2030 1500 800 960 Textblock 5
2823 1497 975 1023 Textblock 6

tesseract-docker.sh

#!/bin/sh
docker run -ti --rm -v "$(pwd):/data" tesseractshadow/tesseract4re /bin/bash -c "tesseract /data/filename.jpeg /data/filename -l deu --psm 4 --oem 3 hocr txt"