/DocAuth

Bajaj | 🌈

Primary LanguagePythonMIT LicenseMIT

DocAuth - Document forgery Detection Application.

Submission for Bajaj Hackathon HackRx 4.0

This is an implementation of a Python script to detect a series of forgeries that can happen in a document. Our basic module supports

  • Signature fraud detection and analysis.
  • Copy and move forgery detection.
  • Identification document forgery detection.
  • Routine document forgery detection and analysis.

[Working]

  1. We have tried implementing a cloud-based web application where the user is asked to choose whether to check signature forgery or document forgery.

  2. Based on his choice he is asked to upload the scanned document. Our machine learning algorithms and neural network-based artificial intelligence detection technique.

  3. The results generate a graph of the analysis and show the areas where the forgery has been done. It shows the percentage accuracy of the report which is classified as follows:

    0% - 10% : Authentic 10% - 55% : Suspicious 55% - 60% : Forged

Attachments

  • The source code of all the Python files used for analysis.
  • The backend and frontend code of the web application.
  • Several analysis results were generated from the code with the discrepancies underlined.
  • Design of the web app.

Source and Papers used

  1. Farid, H., 2009. Seeing is not believing. IEEE Spectrum, 46(8).
  2. Farid, H., 2009. Image forgery detection. IEEE Signal Processing Magazine, 26(2), pp.16-25.
  3. Johnson, M.K. and Farid, H., 2005, August. Exposing digital forgeries by detecting inconsistencies in lighting. In Proceedings of the 7th workshop on Multimedia and security (pp. 1-10). ACM.