pipeline.py

config = {
  "directory": "OCR/Invoices1",
  "ocr_library": "pytesseract",
  "output_file": "results-bc-without.txt",
  "statistics_file": "statistics-bc-without.txt",
  "apply_cropping": False,
  "resize_to_fhd": True,
  "deskew_image": True,
  "threshold_method": "niblack",
  "skip_ocr_processing": False,
  "split_with_commas": False,
  "apply_nlp": False,
  "data_extraction_strategy": "regexp",
  "output_directory": "cropping",
  "zipped_folder": "packed-cropping",
  "save_intermediate_steps": False
}

run_ocr(config)

run_data_extraction(config)

pack_folder_in_parts(config)