Licence Extraction and OCR Demonstration

Quick Start

Requires Python 3.9 or newer.

Required packages are contained in requirements.txt.

pip install -r requirements.txt

Package versions should be flexible in most cases.

Start GUI by:

cd src
python3 App.py

Requires OpenCV.

Directory Structure

• src/
  • App.py: A runnable script which starts the Demonstration GUI.
  • DocumentExtraction/
    • PageExtractor.py: Contains the main class for extracting images licences from images. Doubles as a runnable script for testing the LicenceExtractor.
    • Processors.py: Contains various classes used in Licence Extraction.
    • Constants.py: Contains some constants which control the behaviour of licence extraction.
  • FieldMatching/
    • names/: Contains data used when detecting names in OCR output.
    • FieldExtraction.py: Provides get_matches() which returns the fields that could be extracted from an image.
    • TextractWrapper.py: Largely taken from the AWS Textract documentation. Interface to query Textract.
    • kvp.py: Also mostly from AWS documentation. Retrieves the Key Value Pairs from a Textract response.
• sample-inputs/: Some sample documents for Licence Extraction and OCR.
• output/: The resulting output from Licence Extraction on the sample inputs.
• process-output/: The output from the intermediate steps of Licence Extraction.