OCR++ : A Framework to extract information from Scholarly Articles

NOTE : The tool works only on Linux-based systems

Installation :

• Install the dependencies.

• Clone this repo and copy the OCR++ folder in /var/www/html and make a directory named "media" in /var/www/html

  git clone https://github.com/ocrplusplus/ocrplusplus
  cd ocrplusplus
  mkdir -p /var/www/html/media
  mkdir -p /var/www/html/media/documents
  cp -r OCR++ /var/www/html/ 

Usage:

• With localhost as server :

  • Run the following commands :

    cd /var/www/html/OCR++
    python manage.py runserver

  • Now open a web-browser and go to 127.0.0.1:8000/home

• Without a server :

  • Rename and cpoy the pdf as "input.pdf" on which you want to run OCR++ in the given directory using the command

    cp path/to/pdf /var/www/html/OCR++/myproject/media/documents/input.pdf

  • run the OCR++ engine by running the Script

    cd /var/www/html/OCR++/myproject/media/documents/
    python main_script_batch.py

Dependencies:

- nltk (maxent_pos_tagger + averaged_perceptron_tagger)
- django <=v1.8
- crfpp (Try downloading the source from http://www.filewatcher.com/m/CRF++-0.58.tar.gz.790570-0.html)