Skip to content

Latest commit

 

History

History
36 lines (29 loc) · 2.43 KB

README.md

File metadata and controls

36 lines (29 loc) · 2.43 KB
Raw

Web-OCR

PRs Welcome npm License: MIT

Displays the text in English on an image, using Tesseract-OCR API to process the image to text

  • Users can upload the image from the local system and the text on the image will be displayed in a panel box in the same format as in the image.

Video presentation of the application running here

Assumptions and Limitations

  • Reading the Improve Quality Wiki page for stated to work on Rescaling, Binarisation,Noise Removal,Rotation/Deskewing,Border Removal
  • It fails if the text in the image is rotated upto certain extent and also when there are borders, i.e fails for border removal
  • The accuracy of the text displayed becomes less if the font-size is less, and also if some new font-family is used.

Accuracy and Adaptability

  • Jimp image processing library is used to Rescale the image,converting to black and white, noise in image, rotation upto certain extent.
  • The accuracy isn't 100%, but works fine with clear text on an image.

Running the Project

  • Go through the InstallationGuide.md to install Leptonica and Tesseract-OCR, also install node
  • Head to the directory of the application using Terminal and type npm install to download the packages
  • Type npm start
  • Head to a Browser and type http://localhost:3000/

Contribution

  • The repository is open to contribution from all interested developers.
  • You may go through CONTRIBUTING.md to get detailed instructions on how and where you can contribute.
  • Also, you can write by opening an Issue and also solve a current issue if possible.
  • Fork this project to your Github acoount.
  • After forking, clone the repository to local system and make the necessary changes.
  • Kindly send Pull Requests with explanation as to what changes you have done.

License