Textract

• The main motive behind this project was that we often faced the problem of separately typing any content instead of copy-pasting from an already existing document or image which are not in typed format.

• Hence, a text extractor which would simply scanning and extracting the content of the file would save loads of time and also reduce the chances of typographical error to 0%.

Application Link:

• Desktop Version (For images & PDF's) : http://18.222.220.89:5000/ (currently AWS instance is terminated)
• Mobile Version (Downloadable APK package only for images) : https://drive.google.com/file/d/1JLeMwO4LZjzQcbgoA5ARX-z3L8OsZ4-M/view?usp=sharing

Flow of the Application

• Our system takes the scanned image/document from the user as an input.

• Then perform some image pre-processing techniques, like scaling, binarization and noise removal.

• Use Optical Character Recognition using Tesseract Engine and extract the text.

Usage Guidelines:

1. Desktop Version

• The link http://18.222.220.89:5000/ lands on this page, where you can submit the file from which you want to extract text.

  

• After uploading and submitting a file, the result appear as shown in the image and you click on Copy To Clipboard to copy and use the text as you want.

  

2. Mobile Version

• After downloading the APK package from this link, install it in your device and start the app.

  

• Upload an image and the results come out as follows. Then simply copy-paste the text and use it as per you requirement.

  

Technology Used:

• Flask
• Tesseract OCR Engine
• TensorFlow, OpenCV
• Flutter
• AWS (Deployment)